Track and persist training history per class #47
Merged
Conversation
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
Adds a training_history dictionary to AdaptiveClassifier to track the cumulative number of training examples per class. Updates prediction logic to use training_history for weighting, ensures training_history is saved/loaded with backward compatibility, and includes tests for confidence consistency, continuous learning, backward compatibility, and new class detection. Also adds a GitHub Actions workflow for running tests and coverage.
|
This will fix #44 |
Ensures the FAISS index is rebuilt after new examples are added to maintain up-to-date nearest neighbor search. Test assertions for confidence thresholds are relaxed to account for lower confidence scores due to prototype normalization, improving test robustness.
Bump package version to 0.0.15 in setup.py. Add CHANGELOG.md documenting recent changes and fixes. Update GitHub Actions workflow to test only with Python 3.12 instead of a matrix of versions.
Added tests to verify that prediction confidence remains consistent before and after saving/loading AdaptiveClassifier, especially for single-example-per-class cases. Updated test_memory.py to skip memory efficiency test if psutil is unavailable. Updated GitHub Actions workflow to install psutil for testing.
Label IDs are now assigned in alphabetical order when new classes are added, ensuring consistent mappings regardless of input order. Updated the README to document order dependency in online learning and added comprehensive tests to verify order independence and label assignment behavior.
Removed unnecessary sorting and reordering of input texts in the _get_embeddings method. The function now directly processes the input texts and returns embeddings in the original order, improving code clarity.
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
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.
Adds a training_history dictionary to AdaptiveClassifier to track the cumulative number of training examples per class. Updates prediction logic to use training_history for weighting, ensures training_history is saved/loaded with backward compatibility, and includes tests for confidence consistency, continuous learning, backward compatibility, and new class detection. Also adds a GitHub Actions workflow for running tests and coverage.