Overview

DeepTab is a library for deep learning on tabular data with a scikit-learn compatible interface. It handles feature preprocessing, batching, and the training loop, so the workflow stays fit, predict, and evaluate instead of hand-written PyTorch code and data loaders. Every architecture supports three tasks: classification, regression, and distributional regression for uncertainty quantification.

What is DeepTab?

DeepTab provides 15 stable neural architectures for tabular data:

Family

Models

Notes

State Space Models

Mambular, MambaTab, MambAttention

Mamba-inspired; linear feature-sequence scaling

Transformers

FTTransformer, TabTransformer, SAINT, AutoInt

Feature, row, and self-attention over feature interactions

Residual networks

ResNet, TabR

Skip-connection MLP and retrieval-augmented

Tree-inspired

NODE, ENODE, NDTF

Differentiable soft-tree structures

General baselines

MLP, TabM, TabulaRNN

Dense, parameter-efficient ensemble, and recurrent

Plus 3 experimental models: ModernNCA, Tangos, Trompt

Example:

from deeptab.models import FTTransformerClassifier

model = FTTransformerClassifier()
model.fit(X_train, y_train, max_epochs=100)
predictions = model.predict(X_test)
metrics = model.evaluate(X_test, y_test)

One model, three tasks

Every architecture comes in three variants. Change the suffix to change the task:

Class

Task

Output

*Classifier

Classification

Labels and probabilities

*Regressor

Regression

Continuous values

*LSS

Distributional regression

Distribution parameters

 
from deeptab.models import MambularClassifier, MambularRegressor, MambularLSS

clf = MambularClassifier()   # labels and probabilities
reg = MambularRegressor()    # point estimates
lss = MambularLSS()          # full predictive distribution

The interface is identical across all three, so you can move between tasks, or swap architectures, without rewriting your pipeline.

Design Philosophy

Familiar Interface

If you know scikit-learn, you know DeepTab. Standard fit/predict API with seamless integration:

from sklearn.model_selection import GridSearchCV
from deeptab.models import FTTransformerClassifier

search = GridSearchCV(FTTransformerClassifier(), param_grid, cv=5)
search.fit(X, y)

Defaults and Configuration

With no configuration, DeepTab detects feature types, encodes and scales them, and imputes missing values during preprocessing. Training runs on a GPU when one is available, with early stopping and checkpointing enabled.

Each layer is configurable when you need it. Architecture, preprocessing, and training settings live in separate config objects, so you can change one without touching the others:

from deeptab.configs import ResNetConfig, PreprocessingConfig, TrainerConfig
from deeptab.models import ResNetClassifier

model = ResNetClassifier(
    model_config=ResNetConfig(d_model=128),
    preprocessing_config=PreprocessingConfig(numerical_preprocessing="quantile"),
    trainer_config=TrainerConfig(lr=1e-3, batch_size=256),
)

Built for real datasets

Beyond the defaults above, DeepTab handles details that come up with real data:

  • Mixed numerical, categorical, and precomputed embedding features in a single model

  • Automatic stratified splits for classification, preserving class proportions

  • Mini-batch training that scales to datasets larger than a single batch

  • Multi-device and other Lightning training strategies, enabled by forwarding trainer arguments to fit()

When to Use DeepTab

DeepTab and gradient-boosted trees (XGBoost, LightGBM, CatBoost) are complementary tools. The table below shows where each tends to be the stronger starting point.

Situation

Stronger starting point

Mixed numerical and categorical features

DeepTab or gradient-boosted trees

A few thousand samples or more

DeepTab

Complex feature interactions

DeepTab

Calibrated uncertainty through distributional regression

DeepTab (*LSS variants)

Classification, regression, and distributional models behind one interface

DeepTab

Integration with scikit-learn pipelines and GridSearchCV

DeepTab or gradient-boosted trees

Small datasets where overfitting is a risk

Gradient-boosted trees

Data that does not fit in memory

Gradient-boosted trees

Latency-critical inference

Gradient-boosted trees

Note

These are general guidelines, not strict rules. Results depend on the dataset, so when performance matters it is worth benchmarking DeepTab against a gradient-boosted baseline.

Next Steps