Procuring understandable predictions from the model

[SwagatSBhuyan]

I know this is simply a benchmarking repository. I am trying to extend the model to take in sentences and return the label based on the model. But currently the model only gives the results of the metrics and the code is really complicated as the inputs and predictions are very complex sets of tensors, and I am being unable to attain the required information from these. It would be really kind of you to at least guide me to make the best out of this repository. Thanks!

[dh1105]

Hey! Sure thing, if you have any specific questions feel free to ask me. To execute the notebook, sequential execution of the code blocks should work. Are you encountering a particular error?