Metrics only print for training, not validation

[ncullen93]

Metrics only print for training, not validation

[achaiah]

@nikcheerla Are you certain? Looking at the code it looks like validation metrics are being computed (around line 620) in module_trainer.py:

             if has_validation_data:
                    val_loss = self.evaluate_loader(val_loader,
                                                    cuda_device=cuda_device)
                    if self._has_metrics:
                        val_loss, val_metric_logs = val_loss
                        epoch_logs.update(val_metric_logs)
                    epoch_logs['val_loss'] = val_loss
                    self.history.batch_metrics['val_loss'] = val_loss
                
                # END OF EPOCH

[nikcheerla]

@achaiah Huh, it does look like its computing validation metrics. When I run the basic MNIST example though (at https://github.com/ncullen93/torchsample/blob/master/examples/mnist_example.py) with the exact same code, I don't see any validation metrics printed ...

[recastrodiaz]

@ncullen93 I see that you have recently forced merge to master. Is this still an issue? If so, do you want me to update the PR?

Edit: Just checked, validation metrics are printed in master branch. Close issue?

Edit 2: Scratch that, not sure what I was testing before. This bug is still in master. Will send a PR.