Conversation
There was a problem hiding this comment.
Pull request overview
This PR adapts the codebase to work with an updated n3fit API by migrating from passing individual parameters to _pdfNN_layer_generator to using the new ReplicaSettings class, and correcting a parameter name in write_exportgrid calls.
Key Changes:
- Imports and uses the new
ReplicaSettingsclass fromn3fit.model_gento configure neural network replicas - Updates
_pdfNN_layer_generatorfunction call to acceptreplicas_settingsinstead of individual parameters (nodes,activations,initializer_name,layer_type,seed,impose_sumrule,num_replicas) - Fixes parameter name from
QtoQ0inwrite_exportgridfunction calls
💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.
|
The build test has failed because the right version of |
…x-n3fit-NNgeneration-compat
LucaMantani
left a comment
There was a problem hiding this comment.
Looks good to me! Thanks @ElieHammou
comane
left a comment
There was a problem hiding this comment.
Looks good to me, thanks for fixing it
This PR addresses the issue #24.
I have imported the new
ReplicaSettingsclass and created an instance of it for_pdfNN_layer_generatorto use as an argument.Another issue was caused by the function
write_exportgridfromcolibri.export_resultswhich was expectingQ0as an argument rather thanQ. I have fixed this as well.Note that there is still a mismatch between the previous keys (
initializer_nameandlayer_type), still present in the example runcard, and the new ones expected byReplicaSettings(initializerandlayer_type). It could be nice to harmonise this before merging but it is purely cosmetic:The fixes were successfully test on the example POD runcard.