Dev pytorch

[niekdejonge]

Here we will finalize the integration of pytorch.

To do:

• Implement training models in pytorch. Including reimplementing binning and data generators. Done in Pytorch version (work in progress) #165
• Set random seed for validation data generator. --> Create fully deterministic validation generator #172 (we voted for different strategy!)
• Switch concept of validation step to an all-vs-all computation --> Add more efficient validation function #177
• Integrate the validation data generator into the training pipeline @niekdejonge
• Integrate with the rest of the training workflow and inference. @niekdejonge will do this. Done in Integrate pytorch training in train ms2deepscore #169
• Maybe change the detailed binning to having overlapping second bins. @florian-huber would this be easy for you to do?
• Update MS2DeepScore plus the MC version of the matchms style scoring classes --> Update MS2DeepScore plus MC version for scoring #171
• Remove all old tensorflow functions and refactor old tests we want to keep --> Remove tensorflow parts from dev_pytorch branch. #170
• Reimplement monte carlo dropout for pytorch --> Update MS2DeepScoreMonteCarlo to pytorch #175
• Make TensorizationSettings also an input of train_ms2deepscore and training wrapper
• Actually use the settings in train() as well. (patience, loss function, early_stopping etc. are currently not set

[sonarcloud]

Quality Gate passed

The SonarCloud Quality Gate passed, but some issues were introduced.

18 New issues
0 Security Hotspots
87.5% Coverage on New Code
0.0% Duplication on New Code

See analysis details on SonarCloud