Average training and validation losses using torch.distributed for monitoring trends

[bumi001]

1. Added code for deterministic computation
   • helps in reproducing results
   • expects torch and cuda versions to be the same
2. Added code to calculate average training loss over the entire training dataset across all gpus
   • uses torch.distributed.reduce
   • printed on a per epoch basis
   • helps monitor training trend.
3. Added code to calculate average validation loss over the entire validation dataset across all gpus
   • uses torch.distributed.reduce
   • printed whenever it is calculated
   • helps monitor validation trend.
4. Added code to save the history of average training loss and average validation loss, as needed, as a function of epoch.