tabensemb.model#

AbstractModel(trainer[, program, ...])

The base class for all model bases.

Adaptation for PyTorch#

TorchModel(*args[, lightning_trainer_kwargs])

The class for PyTorch-like models.

AbstractNN(datamodule, **kwargs)

A subclass of pytorch_lightning.LightningModule that is compatible with TorchModel and has implemented training and inferencing steps.

For Supported Model Bases#

AutoGluon(trainer[, program, model_subset, ...])

WideDeep(*args, **kwargs)

PytorchTabular(trainer[, program, ...])

An Example of TorchModel#

CatEmbed(*args[, lightning_trainer_kwargs])

Get Hidden Representations from Models#

AbstractWrapper(model)

For those required deep learning models, this is a wrapper to make them have hidden information like hidden_representation or something else extracted from the forward process.

TorchModelWrapper(model)

WideDeepWrapper(model)

PytorchTabularWrapper(model)

Recursive Feature Elimination#

RFE(trainer, modelbase[, model_subset, ...])