Skip to content

fix: fill inspiration slots in small islands - #508

Merged
codelion merged 1 commit into
algorithmicsuperintelligence:mainfrom
liuzhengyang699:fix/small-island-inspirations
Oct 11, 2026
Merged

codelion merged 1 commit into
algorithmicsuperintelligence:mainfrom
liuzhengyang699:fix/small-island-inspirations

Conversation

@liuzhengyang699

Copy link
Copy Markdown
Contributor

When an island has no more programs than the requested inspiration count, _sample_inspirations skips its diversity and random-fill stage. For example, with a parent and two alternatives, requesting five inspirations returns only the elite instead of both alternatives. This unnecessarily narrows the code examples available to the next generation.

Run the fill stage whenever inspiration slots remain. Existing parent exclusion, uniqueness, elite priority, and island isolation still apply; an undersized island simply contributes its available alternatives.

Validation: new regression cases cover an island equal to and smaller than the requested count, a one-slot request, and a parent-only island. The small-island cases fail before the fix. 16 focused island sampling tests and the repository’s isort/Black pre-commit hooks passed. No model calls were made.

@CLAassistant

CLAassistant commented Oct 10, 2026 •

Copy link
Copy Markdown

CLA assistant check
All committers have signed the CLA.

@CLAassistant

Copy link
Copy Markdown

CLA assistant check
Thank you for your submission! We really appreciate it. Like many open source projects, we ask that you sign our Contributor License Agreement before we can accept your contribution.
You have signed the CLA already but the status is still pending? Let us recheck it.

@codelion
codelion merged commit 8ff5036 into algorithmicsuperintelligence:main Oct 11, 2026
4 checks passed
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.

3 participants