Add autoresearch auto data pruning workflow#2
Draft
ahazeemi wants to merge 4 commits into
Draft
Conversation
Describes how to adapt Karpathy's autoresearch autonomous experiment loop to explore data pruning strategies using dPrune's scorer/pruner pipeline, rather than model architecture changes. https://claude.ai/code/session_01VGVrW3yjH69rVLLqafhjjc
Implements the autonomous experiment loop for data pruning using dPrune, following Karpathy's autoresearch pattern: - prepare.py: Data download, BPE tokenizer, dataloaders, val_bpb eval - prune.py: Mutable dPrune pipeline (agent's target file) - train.py: Minimal GPT with fixed 5-min time budget - program.md: AI agent instructions with search strategy guide - run.sh: Experiment runner with keep-or-revert git logic The agent edits prune.py to explore scorer/pruner/ratio combinations, trains on the pruned subset, and keeps changes that improve val_bpb. https://claude.ai/code/session_01VGVrW3yjH69rVLLqafhjjc
ahazeemi
changed the title
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
2 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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
Summary
Add an
autoresearch/workflow for iterative auto data pruning experiments.What changed
Why
The branch adds a focused harness for exploring data-pruning strategies with a fixed-budget training loop while keeping the workflow runnable on the default TinyStories setup.
Validation
python3 -m compileall autoresearchbash -n autoresearch/run.shNotes
This PR is opened as a draft.