rllm.nn¶

Conv¶

Graph Conv¶

`MessagePassing`	Base class for message passing.
`GCNConv`	The GCN (Graph Convolutional Network) model implementation with message passing, based on the "Semi-supervised Classification with Graph Convolutional Networks" paper.
`LGCConv`	The LGC (Lazy Graph Convolution) implementation with message passing, based on the "From Cluster Assumption to Graph Convolution: Graph-based Semi-Supervised Learning Revisited" paper.
`GATConv`	The GAT (Graph Attention Network) model implementation with message passing, based on the "Graph Attention Networks" paper.
`HANConv`	The Heterogeneous Graph Attention Network (HAN) model implementation with message passing, as introduced in the "Heterogeneous Graph Attention Network" paper.
`HGTConv`	The Heterogeneous Graph Transformer (HGT) layer implementation with message passing, as introduced in the "Heterogeneous Graph Transformer" paper.
`SAGEConv`	Simple SAGEConv layer implementation with message passing, as introduced in the "Inductive Representation Learning on Large Graphs" paper.

Table Conv¶

`ExcelFormerConv`	The ExcelFormerConv Layer introduced in the "ExcelFormer: A neural network surpassing GBDTs on tabular data" paper.
`FTTransformerConv`	The FT-Transformer backbone in the "Revisiting Deep Learning Models for Tabular Data" paper.
`SAINTConv`	The SAINTConv Layer introduced in the "SAINT: Improved Neural Networks for Tabular Data via Row Attention and Contrastive Pre-Training" paper.
`TabTransformerConv`	The TabTransformer LayerConv introduced in the "TabTransformer: Tabular Data Modeling Using Contextual Embeddings" paper.
`TromptConv`	The TromptConv Layer introduced in the "Trompt: Towards a Better Deep Neural Network for Tabular Data" paper.
`TransTabConv`	Single Transformer encoder layer for TransTab ("TransTab").
`ResNetConv`	The ResNet-like TNN LayerConv introduced in the "Revisiting Deep Learning Models for Tabular Data" paper.

Pre-Encoder¶

`FTTransformerPreEncoder`	The FTTransformerPreEncoder class is a specialized pre-encoder for the FTTransformer model.
`TabTransformerPreEncoder`	The TabTransformerEncoder class is a specialized pre-encoder for the TabTransformer model.
`TransTabPreEncoder`	Pre-encoder for TransTab ("TransTab").
`ResNetPreEncoder`	The pre-encoder for ResNet TNN.
`HeteroTemporalEncoder`	HeteroTemporalEncoder for RDL model from paper "RelBench: A Benchmark for Deep Learning on Relational Databases".

Models¶

`RECT_L`	The RECT model, or more specifically its supervised part RECT-L, from the "Network Embedding with Completely-imbalanced Labels" paper.
`BRIDGE`	The BRIDGE model introduced in the "rLLM: Relational Table Learning with LLMs" paper.
`TransTab`	Base TransTab encoder for tabular data ("TransTab").
`TableResNet`	The ResNet-like TNN introduced in the "Revisiting Deep Learning Models for Tabular Data" paper.
`HeteroSAGE`	The heterogeneous version of the GraphSAGE model.
`RDL`	Relational Deep Learning (RDL) model from paper "RelBench: A Benchmark for Deep Learning on Relational Databases".
`RelGNN`	The RelGNN model is a GNN framework specifically designed to leverage the unique structural characteristics of the graphs built from relational databases from paper "RelGNN: Composite Message Passing for Relational Deep Learning".
`RelGNNModel`	The relational table learning model with RelGNN as the HGNN backbone from paper "RelGNN: Composite Message Passing for Relational Deep Learning".

Loss¶

`ContrastiveLoss`	Generalized InfoNCE-style contrastive loss with a customizable positive mask.
`SelfSupervisedVPCL`	The self-supervised vertical-partition contrastive loss (Self-VPCL) implementation, based on the "TransTab: Learning Transferable Tabular Transformers Across Tables" paper.
`SupervisedVPCL`	The supervised vertical-partition contrastive loss (Supervised-VPCL) implementation, based on the "TransTab: Learning Transferable Tabular Transformers Across Tables" paper.