Repository navigation
[tmva] Deprecate the deep learning methods and their build options - #24255
Merged
Merged
Conversation
MethodDL, MethodDNN and the TMVA::DNN classes are deprecated and will be removed in ROOT 6.46. Neural networks are trained with dedicated machine learning libraries nowadays, so maintaining a separate implementation with CPU, CUDA and cuDNN backends is no longer justified. The `tmva-cpu`, `tmva-gpu`, `tmva-cudnn` and `use_gsl_cblas` build options only affect these methods and are deprecated with them. The `tmva-cpu` option is now OFF by default, also with `all=ON`, so that only builds that ask for it get the configure-time warning. The `tmva-cudnn` option stays ON by default, but it only has an effect together with `tmva-gpu`, so it only warns in that case. Both methods also warn at runtime, because users of pre-built binaries never see the configure-time warning. Assisted-by: claude-code:claude-fable-5-2
The TMVA deep learning methods are deprecated, so the tutorials should
not promote them anymore:
* TMVA_CNN_Classification and TMVA_RNN_Classification only make sense
with these methods and are removed, together with the helper
PyTorch_Generate_CNN_Model.py that nothing else uses.
* The other TMVA tutorials no longer book the deep learning methods.
TMVA_Higgs_Classification moves to the basic training tutorials in
the index.
TMVAMulticlass and TMVA_Higgs_Classification are single-threaded now,
so they leave the list of multithreaded tutorial tests.
TMVA_Higgs_Classification.py is also dropped from the tutorials that
require TensorFlow, which it has not used since the removal of PyMVA,
and loses an unused import that the ruff CI job rejects.
Assisted-by: claude-code:claude-fable-5-2
guitargeek
requested review from
dpiparo,
lmoneta and
silverweed
as code owners
November 19, 2026 16:17
dpiparo
approved these changes
Nov 19, 2026
Test Results 22 files 22 suites 3d 9h 50m 9s ⏱️ Results for commit a515e27. |
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
MethodDL, MethodDNN and the TMVA::DNN classes are deprecated and will be removed in ROOT 6.46. Neural networks are trained with dedicated machine learning libraries nowadays, so maintaining a separate implementation with CPU, CUDA and cuDNN backends is no longer justified.
The
tmva-cpu,tmva-gpu,tmva-cudnnanduse_gsl_cblasbuild options only affect these methods and are deprecated with them. Thetmva-cpuoption is now OFF by default, also withall=ON, so that only builds that ask for it get the configure-time warning. Thetmva-cudnnoption stays ON by default, but it only has an effect together withtmva-gpu, so it only warns in that case.Both methods also warn at runtime, because users of pre-built binaries never see the configure-time warning.
Note that MethodDNN has actually been deprecated already 5 years ago in c600414, but just had no clear removal path.