tabular-deep-learning

DeepTab is a Python package that simplifies tabular deep learning by providing a suite of models for regression, classification, and distributional regression tasks. It includes models such as Mambular, TabM, FT-Transformer, TabulaRNN, TabTransformer, and tabular ResNets.

Awesome Tabular Deep Learning for "Representation Learning for Tabular Data: A Comprehensive Survey"

Open-source tabular foundation model for fast, training-free regression with predictive distributions.

GOTabPFN: Graph-guided feature ordering and compact tokenization for TabPFN-style tabular foundation models on high-dimensional data.

Official repository for ICML2026 "TopBench: A Benchmark for Implicit Predictive Reasoning in Tabular Question Answering"

⚡ Universal Tabular Auto-ML & Deep Learning Engine in PyTorch, Optuna & Gradio

A curated collection of TDL (Tabular Deep Learning) resources—libraries, projects, tutorials, papers, and more—for researchers and developers in the field.

PyTorch implementation of the FT-Transformer (Gorishniy et al., 2021) for tabular regression — benchmarked against classical baselines on California Housing, with ablation studies, Optuna tuning, 98% test coverage, and CI/CD.

Code accompanying our TMLR paper, benchmarking B-spline, M-spline, I-spline, and PLE numerical encodings for tabular deep learning, including fixed, target-aware, and learnable knot placement.

This study proposes a diverse family of additive models that extend Generalized Additive Models to achieve both interpretability and predictive performance in forecasting. While preserving the stability of traditional GAMs, we develop Feature-wise Additive Models, Gradient Boosting Additive Models, and TabNet-based Neural Additive Models (TabNAM).

Tabular deep learning and Streamlit deployment for cervical high-grade lesion risk prediction.

Code for in-progress paper "Deep Learning for Credit Risk: Evaluating Feature Tokenizer Transformers Against Industry Baselines"

Quick overview of ML & deep learning foundations for fraud-detection: classical models, DL, TF/PyTorch, and deployment patterns.