6
$\begingroup$

Are there any tools to monitor network's training in PyTorch?

I am looking for an equivalent to tensorboard in tensorflow.

$\endgroup$
4
$\begingroup$

I am using tensorboardX. It supports most of the features of TensorBoard. I am using the Scalar, Images, Distributions, Histograms and Text. Haven't tried the rest like audio and graph. But the repo also contains examples for those usecases. Installation can be done easily with pip. It's all explained in the readme.

There are also other software which implement a wrapper for PyTorch (and other languages/frameworks) of TensorBoard. As far as I know, they support fewer functionalities. But have a look at Crayon and Tensorboard-Logger.

$\endgroup$
2
$\begingroup$

I am also using Tensorboard-PyTorch (TensorboardX). Over all I am quite happy with it.

But don't try to visualize graphs. At least none with a bit of complexity (e.g. a resnet50 won't work). There are some issues about it on their github page.

But better check out the Pytorch forum frequently. Pytorch seems to move quite fast. And a direct tensorboard implementation is planned, as well.

$\endgroup$
1
$\begingroup$

Yes, there are a few tools: TNT, tensorboardX and VisualDL. Have a look at this blog post https://nbro.github.io/blogging/2019/01/06/an-example-of-how-to-use-visualdl-with-pytorch/, where an example of how to use VisualDL is given, for more details.

$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.