Explainability framework SHAP main code, masking of input Issue

[Kathleendeboer]

Code for issue #15

The code contains:

• Loading the model
• Gathering the data
• Generating tokens representing the input
• Masking the input based on the masked tokens
• Running the model on all combinations of masked input
• Storing the predictions and probabilities

This code still needs:

• Testing of the inference of the model on the masked combinations
• Connection to input from Sound and Vision:
  • It now uses the validation mode from the model which needs target labels, maybe add dummy labels?
  • Connect to pre process module (in progress) Issue #2

[Kathleendeboer]

Recently pushed running code for explainable emotion recognition with segmentation. I provided a test notebook for running an example. #2 #16

[Veldhoen]

Your pyproject.toml seems to be invalid. Can you fix it to make sure your project can be installed and has all the proper dependencies? Or, at the minimum, provide a requirements.txt Also, I got some indentation errors upon importing your data preprocessing module: something weird seems to have happened with your docstrings. And please indicate what files to put where in order to run your test (model checkpoint, pretrained models, input files..)