Add reporting features on training (and validation) results
At the moment, we only compute metrics at evaluation time, whereas it would be nice to also report on the training loss.
It would also be nice to enable visualizing results (whether from training or evaluation rounds) via GUI tools such as TensorBoard and/or Weights & Biases. This may either be done by integrating calls to such a tool within our code, or by writing post-processing utils to input data from our exported csv/json/log file(s) into the tool.