Add 14-class merged-group labels and no-re-extract relabel tool for l…#3
Merged
Conversation
…inear probing - scripts/eval_labels/splits_merged_majority/: the 14 clinical-group labels for the probe. Each of 37 pathologies is set by a 3-model majority vote (Claude Opus 4.7 + GPT-5.5 + Nemotron-3 Super 120B), then collapsed into the neuroradiologist's 8 pathophysiology (PP_*) + 6 imaging-phenotype (BP_*) groups. Reproducible via build_merged_group_labels.py --source majority (the builder also supports raw/csv32 agreement variants on demand). - scripts/relabel_features.py: build a linear-probe feature dir for a different labels CSV by reusing already-cached encoder features (symlink features_*.npy / subject_ids_*.txt, rebuild only labels_*.npy + label_names.json). Avoids re-running the frozen encoder when only labels change; errors out if a cached subject is missing from the new labels CSV. - README: Linear Probing section documents the label set (class count is derived from the CSV, no code change) and the extract-once / relabel workflow. Bulky derived CSVs (all/train/val/test) are gitignored.
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
1 participant
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
…inear probing
scripts/eval_labels/splits_merged_majority/: the 14 clinical-group labels for the probe. Each of 37 pathologies is set by a 3-model majority vote (Claude Opus 4.7 + GPT-5.5 + Nemotron-3 Super 120B), then collapsed into the neuroradiologist's 8 pathophysiology (PP_) + 6 imaging-phenotype (BP_) groups. Reproducible via build_merged_group_labels.py --source majority (the builder also supports raw/csv32 agreement variants on demand).
scripts/relabel_features.py: build a linear-probe feature dir for a different labels CSV by reusing already-cached encoder features (symlink features_.npy / subject_ids_.txt, rebuild only labels_*.npy + label_names.json). Avoids re-running the frozen encoder when only labels change; errors out if a cached subject is missing from the new labels CSV.
README: Linear Probing section documents the label set (class count is derived from the CSV, no code change) and the extract-once / relabel workflow.
Bulky derived CSVs (all/train/val/test) are gitignored.