teras.models module

Backbones

FTTransformerBackbone(input_dim, ...[, ...])	FT-Transformer Encoder backbone based on the FT-Transformer architecture proposed in the "Revisiting Deep Learning Models for Tabular Data" paper.
SAINTBackbone(input_dim, cardinalities, ...)	SAINT Backbone based on the SAINT architecture proposed in the paper, "SAINT: Improved Neural Networks for Tabular Data".
TabNetEncoderBackbone(input_dim, ...[, ...])	TabNetEncoder proposed by Arik et al. in the "TabNet: Attentive Interpretable Tabular Learning" paper.
TabTransformerBackbone(input_dim, ...[, ...])	TabTransformer backbone based on the architecture proposed in the "TabTransformer: Tabular Data Modeling Using Contextual Embeddings".
TransformerEncoderBackbone(input_dim, ...[, ...])	Transformer Encoder model as proposed in the "Attention is all you need" paper.

Pretrainers

SAINTPretrainer(model, cardinalities, ...[, ...])	SAINTPretrainer as proposed in the paper, "SAINT: Improved Neural Networks for Tabular Data".
TabNetPretrainer(encoder, decoder[, ...])	TabNetPretrainer for pretraining TabNetEncoder as proposed by Arik et al. in the paper, "TabNet: Attentive Interpretable Tabular Learning".
TabNetDecoder(data_dim, ...[, ...])	TabNetDecoder model for self-supervised learning, proposed by Arik et al. in the "TabNet: Attentive Interpretable Tabular Learning" paper.
TabTransformerRTDPretrainer(model, data_dim)	Replaced Token Detection (RTD) based Pretrainer for pretraining TabTransformerBackbone as proposed by Huang et al. in the paper, "TabTransformer: Tabular Data Modeling Using Contextual Embeddings".
TabTransformerMLMPretrainer(model, data_dim)	Masked Language Modelling (MLM) based Pretrainer for pretraining TabTransformerBackbone as proposed by Huang et al. in the paper, "TabTransformer: Tabular Data Modeling Using Contextual Embeddings".

Generative

CTGAN(generator, discriminator, metadata[, ...])	CTGAN is a state-of-the-art tabular data generation architecture proposed by Lei Xu et al. in the paper, "Modeling Tabular data using Conditional GAN".
CTGANGenerator(data_dim, metadata[, ...])	CTGANGenerator for CTGAN architecture as proposed by Lei Xu et al. in the paper, "Modeling Tabular data using Conditional GAN".
CTGANDiscriminator([hidden_dims, ...])	CTGANDiscriminator for CTGAN architecture as proposed by Lei Xu et al. in the paper, "Modeling Tabular data using Conditional GAN".
GAIN(generator, discriminator[, hint_rate, ...])	GAIN is a missing data imputation model based on GANs.
GAINGenerator(data_dim[, hidden_dims, ...])	Generator model for the GAIN architecture proposed by Jinsung Yoon et al. in the paper "GAIN: Missing Data Imputation using Generative Adversarial Nets.".
GAINDiscriminator(data_dim[, hidden_dims, ...])	Discriminator model for the GAIN architecture proposed by Jinsung Yoon et al. in the paper GAIN: Missing Data Imputation using Generative Adversarial Nets.
PCGAIN(generator, discriminator, classifier)	PCGAIN is a missing data imputation model based on the GAIN architecture.
TVAE(encoder, decoder, metadata, data_dim[, ...])	TVAE is a tabular data generation architecture proposed by Lei Xu et al. in the paper, "Modeling Tabular data using Conditional GAN".
TVAEEncoder([latent_dim, compression_dims])	Encoder for the TVAE model as proposed by Lei Xu et al. in the paper, "Modeling Tabular data using Conditional GAN".
TVAEDecoder(data_dim[, decompression_dims])	Encoder for the TVAE model as proposed by Lei Xu et al. in the paper, "Modeling Tabular data using Conditional GAN".

Task Models

Classifier(backbone, num_classes[, ...])	Classifier class that provides a dense prediction head.
Regressor(backbone, num_outputs[, ...])	Regressor class that provides a dense prediction head.