Changes in version 0.1.5 (2026-04-23)                  

  - Initial tests added; cleanup for R CMD check portability.
  - Initial CRAN submission.

  - All torch-dependent examples are now guarded by torch_available() to
    ensure safe execution when torch or libtorch is not available.

                 Changes in version 0.1.4 (2026-01-12)                  

  - Added relu to input-skip layers. Updated experiments.

                 Changes in version 0.1.3 (2026-01-07)                  

  - Changed some function names to be consistent with CRAN guidelines.
  - Removed export of function that is only used internally.
  - Changed assignment operator to be consistent everywhere.
  - Removed datasets not used in RJ article submission.

                 Changes in version 0.1.2 (2025-12-10)                  

  - Fix bug in LBBNN_Linear so layers work properly in convolutional
    architecture.
  - Updated experiment on convolutional architecture for R journal
    submission.
  - Added default values to mpm and draws in predict() function.