Skip to content

PWGHF: Implement B0 ML and templatize task - #4661

Merged
fgrosa merged 5 commits into
AliceO2Group:masterfrom
AlexBigO:dev_bmesons
Feb 12, 2024
Merged

PWGHF: Implement B0 ML and templatize task#4661
fgrosa merged 5 commits into
AliceO2Group:masterfrom
AlexBigO:dev_bmesons

Conversation

@AlexBigO

Copy link
Copy Markdown
Contributor

Hi @fgrosa , I implemented the ML inference for B0 in the reduced workflow.

Modifs:

  • new ML-dedicated header for B0
  • reduced B0 selector adapted so one can use ML selection
  • reduced B0 task templatized to process data/MC and also to process either with ML information on D meson or on B0 (if we want D meson ML scores, it is to train BDT for B0; and if we want B0 ML scores, we already used the D meson ML scores for the training so we don't need to access them again --> no need to access D meson and B0 ML scores simultaneously).
  • a bug was fixed: the B0 candidate tables consumed by the MC process functions did not use to be filtered, they are now

@apalasciano I tag you in case you're interested in using ML for B+ :)

@AlexBigO

Copy link
Copy Markdown
ContributorAuthor

Hi, it looks like I can't pass both Clang-format and MegaLinter simultaneously (because of a macro defined in HfMlResponseB0ToDPi.h). Does someone know how to overcome this, please?

@fgrosafgrosa left a comment

Copy link
Copy Markdown
Collaborator

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Thanks a lot @AlexBigO!
I essentially just have a comment about the dimensionality of the output score (since in binary classification signal score = 1 - background score I would only store and use one value)

Regarding the clang format vs Megalinter issue, we had exactly the same for the Lc->pKpi channel with @mfaggin and in the end we merged it with Megalinter red, but maybe @vkucera has a solution for that.

Comment threadPWGHF/D2H/TableProducer/candidateSelectorB0ToDPiReduced.cxx Outdated
Comment threadPWGHF/D2H/TableProducer/candidateSelectorB0ToDPiReduced.cxx
Comment threadPWGHF/D2H/TableProducer/candidateSelectorB0ToDPiReduced.cxx
Comment threadPWGHF/DataModel/CandidateSelectionTables.h Outdated
Comment threadPWGHF/D2H/Tasks/taskB0Reduced.cxx Outdated
Comment threadPWGHF/D2H/Tasks/taskB0Reduced.cxx Outdated
Comment threadPWGHF/D2H/Tasks/taskB0Reduced.cxx Outdated
Comment threadPWGHF/D2H/Tasks/taskB0Reduced.cxx Outdated
Comment threadPWGHF/D2H/Tasks/taskB0Reduced.cxx Outdated
Comment threadPWGHF/D2H/Tasks/taskB0Reduced.cxx
@vkucera

Copy link
Copy Markdown
Collaborator

Hi, it looks like I can't pass both Clang-format and MegaLinter simultaneously (because of a macro defined in HfMlResponseB0ToDPi.h). Does someone know how to overcome this, please?

This seems to be coming from a conflict between the different clang-format versions used in the two CI checks. For now, I would suggest to fix the formatting according to the clang-format check (which is required for the merging) and merge the PR despite the failing MegaLinter check.

@vkuceravkucera closed this Feb 12, 2024
@vkuceravkucera reopened this Feb 12, 2024
@AlexBigO

Copy link
Copy Markdown
ContributorAuthor

Hi, it looks like I can't pass both Clang-format and MegaLinter simultaneously (because of a macro defined in HfMlResponseB0ToDPi.h). Does someone know how to overcome this, please?

This seems to be coming from a conflict between the different clang-format versions used in the two CI checks. For now, I would suggest to fix the formatting according to the clang-format check (which is required for the merging) and merge the PR despite the failing MegaLinter check.

Hi Vit, I will do this then. Thanks!

@fgrosa
fgrosa merged commit 6c908cc into AliceO2Group:masterFeb 12, 2024
basiach pushed a commit to basiach/O2Physics that referenced this pull request Feb 23, 2024
* Implement B0 ML and templatize task
* Solve MegaLinter issue
* Solve MegaLinter without running clang format
* Add missing lines to fill B0 ML scores histograms in processDataWithB0Ml
* Apply Fabrizio's comments
---------
Co-authored-by: Alexandre Bigot <abigot@sbgat402.in2p3.fr>
Sign up for freeto join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Development

Successfully merging this pull request may close these issues.

3 participants

@AlexBigO@vkucera@fgrosa