- feat!: removed data preparation out of predictor class, sub_seq, steps_past, steps_future need now to be defined in each model method - allows for advanced hyper parameter tuning - fix: removed tensor board activation logic bug
2.0.1
- fix: removed dead pandas imports - chore: added tensorflow as base requirement
2.0.0
- adapted `keras_core` - removed internal hyperparameter tuning - removed encoder-decoder architectures - improved layer configuration via dictionary input - split data argument into target and feature numpy arrays
1.3.0
- added depth parameter to architectures - added optimizer configuration support - added optimizer configuration to seeker
1.2.0
- added Tensorboard support - changed show_performance plot to show loss and metric values - added optional dropout and regularization layers to architectures
1.1.0
- removed batch_size parameter from fit_model method - hyperparameter optimization added via the Optuna library