Auto-PyTorch

What Auto-PyTorch is

Auto-PyTorch is a free, self-hosted machine-learning toolkit based on PyTorch. It automates model selection and tuning, including neural-network architecture and training hyperparameters, within a user-set search budget. Its search methods include Bayesian optimization, meta-learning, SMAC, and Hyperband. Supported tasks include tabular classification, tabular regression, and time-series forecasting. For tabular work, preprocessing can include imputation, categorical encoding, scaling, and feature processing, with related hyperparameters tuned during search. Auto-PyTorch can build ensembles based on model predictions for a validation set; users can configure ensemble size and candidate limits. Memory limits for estimators and an overall wall-time limit are configurable. Parallel Bayesian optimization is available through Dask.distributed, but workers need shared filesystem access to training data and models. Installation documentation specifies Linux, Python 3.7 or later, a C++11-capable compiler, and SWIG 3.0; SWIG 4.0 or later is not supported. Forecasting needs additional dependencies. The project provides a Docker image and states that it uses the 3-clause BSD license.