1.2405.0

New Features

• Dependencies: Added huggingface-hub to dependencies.
• Docker Support: Introduced a new Dockerfile for building a Docker image with Python 3.12-slim-bookworm.

Improvements

• Code Formatting: Improved code formatting across various files for better readability and consistency.
• Configuration Management:
  • Moved lint configuration files (.flake8, .style.yapf) to the root directory without changes.
  • Added .mypy.ini and .pylintrc configuration files for type checking and linting.
• Logging: Enhanced logging configuration to set logging levels for various components.
• Model Aggregation: Updated AggregationStrategies with improved parameter handling for binary classification strategies.
• Tokenizer Handling: Improved tokenizer handling and feature extraction in helper.py.
• Split Strategies: Enhanced split strategies to handle grouped sentences and remove short groups more effectively.

Bug Fixes

• Type Handling: Replaced deprecated np.int with int to avoid compatibility issues.
• GPU Name Retrieval: Fixed an issue with GPU name retrieval in __init__.py.

Documentation

• README:
  • Updated download badges for improved display and accuracy.
  • Removed deprecated model names and added a section for HuggingFace models.
• Examples: Updated examples/binary_classifier.py to reflect the latest API changes and improve clarity.

Miscellaneous

• Version Bump: Updated version from 1.2308.0 to 1.2405.0 to reflect new changes and improvements.
• Setup: Modified setup.py to include the new dependencies and ensure compatibility with the latest versions of required packages.

1.2308.0

• Fulfill with zeroes when token_type_ids is missing

1.2307.0

• Add support for Hugging Face Transformers 4.x
• Upgrade dependencies

1.0.1

• Add random_state option

1.0.0

Merge pull request #21 from labteral/fix-install

1.0.0