Adding Trainer

[captain-pool]

Adding Trainer for Knowledge Distillation.

Addresses #42, #45 , #46 , #47, #44

[captain-pool]

Have we added any README or documentation that references the paper we are using for this? And what changes we are making to the loss function, etc?

We have 2 options for Loss. We can choose any of these, depending on the output we are getting.

• Comparative Loss -MSE Loss (As Stated by the Paper).
• [Preferred] Adversarial Loss - Joint MSE and Relativistic Average Loss.

I've implemented both, we will be using each of them based on the output we get.