General

python-Chebai employs deep neural network models for the semantic classification of chemical entities. Model classes are located within the chebai/models/ directory.

• chebai/models/base.py: Contains the base class ChebaiBaseNet, inherited from the PyTorch Lightning module, facilitating custom model creation.

• Example: chebai/models/electra.py showcases a custom model inherited from ChebaiBaseNet, implementing an Electra model.

Regression vs Classificaiton

By default, python-chebai models assume a classification task. However, by setting the parameter model_type to regression in the model yaml file, this can be changed to also consider regression tasks.