bnns 1.0.0
CRAN release: 2026-06-07
-
New Feature: Added full support for the
tidymodelsecosystem, including integration withparsnip,recipes,workflows, andtune. -
New Feature: Added GPU acceleration support for fitting Bayesian Neural Networks via OpenCL and
cmdstanr. -
New Feature: Introduced model evaluation metrics using Leave-One-Out (LOO) cross-validation (
loo.bnns()) and Watanabe-Akaike Information Criterion (waic.bnns()). -
New Feature: Added
save_bnns()andload_bnns()functions for easy saving and loading of trained models. -
New Feature: Added a
plot.bnns()method for visualizing posterior distributions and model convergence. -
New Feature: Added option to provide character input for
act_fnandout_act_fnarguments. - Changed the package to stable lifecycle and prepared for CRAN submission.
- Removed
cmdstanrfrom package dependencies. It is now accessed dynamically to comply with CRAN policies. - Removed the manual rhub GitHub Actions badge from the README.
- Added
CODE_OF_CONDUCT.md.
