cito: Building and Training Neural Networks

The 'cito' package provides a user-friendly interface for training and interpreting deep neural networks (DNN). 'cito' simplifies the fitting of DNNs by supporting the familiar formula syntax, hyperparameter tuning under cross-validation, and helps to detect and handle convergence problems. DNNs can be trained on CPU, GPU and MacOS GPUs. In addition, 'cito' has many downstream functionalities such as various explainable AI (xAI) metrics (e.g. variable importance, partial dependence plots, accumulated local effect plots, and effect estimates) to interpret trained DNNs. 'cito' optionally provides confidence intervals (and p-values) for all xAI metrics and predictions. At the same time, 'cito' is computationally efficient because it is based on the deep learning framework 'torch'. The 'torch' package is native to R, so no Python installation or other API is required for this package.

Version: 1.1
Depends: R (≥ 3.5)
Imports: coro, checkmate, torch, gridExtra, parabar, abind, progress, cli, torchvision, tibble, lme4
Suggests: spelling, rmarkdown, testthat, plotly, ggraph, igraph, stats, ggplot2, knitr
Published: 2024-03-18
DOI: 10.32614/CRAN.package.cito
Author: Christian Amesöder [aut], Maximilian Pichler ORCID iD [aut, cre], Florian Hartig ORCID iD [ctb], Armin Schenk [ctb]
Maintainer: Maximilian Pichler <maximilian.pichler at>
License: GPL (≥ 3)
NeedsCompilation: no
Language: en-US
Citation: cito citation info
Materials: README NEWS
CRAN checks: cito results


Reference manual: cito.pdf
Vignettes: Introduction to cito
Training neural networks
Example: (Multi-) Species distribution models with cito
Advanced: Custom loss functions and prediction intervals


Package source: cito_1.1.tar.gz
Windows binaries: r-devel:, r-release:, r-oldrel:
macOS binaries: r-release (arm64): cito_1.1.tgz, r-oldrel (arm64): cito_1.1.tgz, r-release (x86_64): cito_1.1.tgz, r-oldrel (x86_64): cito_1.1.tgz
Old sources: cito archive


Please use the canonical form to link to this page.