feat(scviva): add auto-calibrated spatial contiguity loss - #36
feat(scviva): add auto-calibrated spatial contiguity loss#36ori-kron-wis wants to merge 1 commit into
Conversation
|
Check out this pull request on See visual diffs & provide feedback on Jupyter Notebooks. Powered by ReviewNB |
![chatgpt-codex-connector[bot]](https://avatars.githubusercontent.com/in/1144995?s=60&v=4){kind=small}
There was a problem hiding this comment.
💡 Codex Review
Here are some automated review suggestions for this pull request.
Reviewed commit: 6f609dcf8d
ℹ️ About Codex in GitHub
Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you
- Open a pull request for review
- Mark a draft as ready
- Comment "@codex review".
If Codex has suggestions, it will comment; otherwise it will react with 👍.
Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".
| if multiple_devices or merged_trainer_kwargs.get("strategy") not in ( | ||
| None, | ||
| "auto", | ||
| ): |
There was a problem hiding this comment.
When num_nodes > 1 is supplied through trainer_config or **trainer_kwargs while devices and strategy remain at their accepted defaults, this guard allows distributed training despite the stated single-device restriction. The custom contiguity loader has no rank-specific edge sampling, so workers can repeatedly regularize the same sampled edges and produce behavior that differs from the supported single-device objective; validate num_nodes here as well.
Useful? React with 👍 / 👎.
1 participant
Description
Brief description of changes and motivation.
Type of Change
Testing
pytest tests/ -v)Checklist
ruff check src/ tests/)