v3.1.0

use mgk_cache_data replace mgk_read_data to efficiently store the graph and features objects.

v3.0.0

The dataset object is simplified and reconstructed. The computation is more efficient for reading large datasets.
Mixture_columns and reaction columns are deprecated.
The entire package is updated accordingly.

v2.3.0

Support fpSize for layered fingerprints.
Support atomInvariantsGenerator for Morgan Fingerprints.

v2.2.0

Support a series of new molecular fingerprints.

v2.1.1

Support in code data augmentation for multiple smiles input.
Support use stored hyperparameters for RBF and DotProduct kernel.
Support color visualization for iterpretable MGK.
Add typed-argument-parser and ipython into the required packages.
Update notebook files.

v2.0.1

Bugfix for --features_hyperparameters_file.
Bugfix for running job on a machine without GPU.
Add feature: Support PreComputed_PureGraph_FeaturesAddMol.

v2.0.0

Reconstruct the code for kernel and kernel config.
Support Optuna for hyperparameter optimization.

v1.2.0

Add executable commands for easy usage.
Support hyperparameter optimization using multiple datasets.

v1.1.1

Bugfix: the predicted uncertainty is not outputed for external test set.

v1.1.0

Add 3 metrics for regression tasks: Spearman's r, Pearson's r, and Kendall's tau.