How to visualize the log files generated by TensorboardX?


I used TensorBoardX to log my losses and accuracy values. I followed their demo given on their github page to save my log file.

I initialized the SummaryWriter() using writer = SummaryWriter(log_dir=log_path). Then in every 10th iteration I was saving the values using: writer.add_scalar('train', loss_avg, iteration_number)

I now have a .json file in the log directory that I provided with no way to visualize it?

How do I actually plot these values?


You can use tensorboard.

From the command line:

$ tensorboard --logdir ./mylogs