Changelog

Version 1.1.3 (Current)

New Features:

• Enhanced D1/D2 layer architecture with improved data handling

• Added support for temporal enrichment (hour, dow, month, minute)

• Implemented global vs local forecasting modes

• Added memory-efficient data loading options

• Improved scaling with separate feature and target scaling

Improvements:

• Modernized Sphinx documentation setup

• Added comprehensive API documentation

• Improved test coverage

• Enhanced PyTorch Lightning integration

• Optimized data loading performance

Bug Fixes:

• Fixed data leakage in scaling operations

• Corrected group_id handling in batch structure

• Fixed temporal cardinality computation

Version 1.1.1

• Initial public release

• Support for multiple state-of-the-art models

• PyTorch Lightning integration

• Basic documentation

Future Plans

• Additional model architectures

• Enhanced visualization tools

• Improved hyperparameter tuning

• Extended benchmark dataset support