feat: add autoresearch skill with cross-platform support and research team integration - #34
Merged
Merged
Conversation
… team integration Implements an autonomous ML experiment loop skill inspired by Karpathy's autoresearch and the macOS fork by miolini. Key design decisions: - User configures all key parameters: training time budget per experiment, metric to optimize (+ direction), max iterations, and allowed modification scope — nothing is hardcoded - Full cross-platform support: CUDA (Linux/Windows), Apple Silicon MPS (macOS M1–M4), and CPU fallback with per-platform adjustments (torch.compile, FlashAttention vs SDPA, batch size guidance, optimizer state casting) - Autonomous loop: read codebase → propose one change → commit → train for T minutes → evaluate metric → keep improvements / discard regressions → log to results.tsv → repeat - Integrated with research team: receives configs from lead-researcher or experiment-design; delivers results.tsv + autoresearch-summary.md to principal-scientist and research-writing teammates - Added to research-team/config.json as a task-layer teammate reporting to lead-researcher - Added to .claude-plugin/marketplace.json under Research category - Updated README.md: added autoresearch to the hierarchy diagram, skill table, and repository structure listing https://claude.ai/code/session_019Zi52EB5qWUbjF3kU4zkK7
adityak74
approved these changes
Mar 8, 2026
adityak74
approved these changes
Mar 8, 2026
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
3 participants
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Implements an autonomous ML experiment loop skill inspired by Karpathy's autoresearch
and the macOS fork by miolini. Key design decisions:
optimize (+ direction), max iterations, and allowed modification scope — nothing is
hardcoded
and CPU fallback with per-platform adjustments (torch.compile, FlashAttention vs SDPA,
batch size guidance, optimizer state casting)
evaluate metric → keep improvements / discard regressions → log to results.tsv → repeat
experiment-design; delivers results.tsv + autoresearch-summary.md to principal-scientist
and research-writing teammates
repository structure listing
https://claude.ai/code/session_019Zi52EB5qWUbjF3kU4zkK7