Dnn and transformer modules! - #4
Merged
Merged
Conversation
…irectly integrate with it such as PrettyPrinting.py and Plotting.py and similarly have updated the DataScaler.py to add my own custom scaler. I have written quite general code that could be integrated with the other modules. I have not changed any code that interacts with the other data modules hence they should run. The only addition I have done (which has no functional change to your ML modules) is add a get_features to the already present scaler classes, something which my RegressionDNN needs to pretty print data but doesn't have any effects when your modules call their scalers.
…scaler (eg. inverse function) and starting on new Transformer module, created general architecture.py to abstract away model construction
…txt file to log key info about the model (eg. Flops and Max Memory used), added same file but .yaml to ease future data extraction and analysis, increased customization from config.yaml of the transformer model (batchnorm, layernorm, attention pooling). Added new needed imports to requirements.txt
…s that are obtained from the config.yaml file, not complete, some features are Transformer only but should be extended to DNN in the fture aswell such as KDE
…Full capabilities are only ecorporated in the PyTorchLightning module (like invserse data sampling and pretty and professional plotting
…ull capabilities are only ecorporated in the PyTorchLightning module (like invserse data sampling and pretty and professional plotting
…They do shared code between themseleves so idea of creating a centralized library to interact from the notebooks is on the table
…compare them head to head, this should probably be in the directory above as the models compared here are more likely trained with torch lightning but could also be trained with the notebooks. This is not fully polished yet
…ner model architecture, bug with predictions still needs to be fixed
…y own regression models. Changes had to be made because 1: my model is created in the setup function which is called in the datamodule, hence the datamodule is created first by hand. 2: when using custom attention layers the weights saved in the checkpoint can have extra names preceding the saved weights such as model.layer_0.attn_head0, this is taken care of from the new function in utils.py, it should not interfere with Diego's models because if a perfect match is found between the weights and the model keys, that is used
… fixed 40 and 1500 GeV
fixed CI bug where sub-directories in results folder were not deleted
diegobaronm
reviewed
Sep 15, 2025
addition of report
diegobaronm
reviewed
Sep 16, 2025
diegobaronm
reviewed
Sep 16, 2025
diegobaronm
reviewed
Sep 16, 2025
diegobaronm
reviewed
Sep 16, 2025
diegobaronm
reviewed
Sep 16, 2025
diegobaronm
reviewed
Sep 16, 2025
diegobaronm
reviewed
Sep 16, 2025
Owner
|
Thanks @FedericoSaitta |
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
2 participants
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.
added my own dnn and transformer regression models along with all their helper modules and a read me file. added the correct testing scripts to the CI and made small modifications to predict, train and performance py and hence to the setup function of your DNN and transformer modules. These changes are minor and are needed for my models as unlike yours, the actual neural netword architecture needs the datamodule to be fully set up to be initialized.