MultiGPU support for cosmoGAN #1
Open
Conversation
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
End goal is multiGPU support via batch-splitting for data parallelization.
In the midst of understanding the code and averaging the split gradients so that the generator and discriminator learn the same things. On the first round of training, some of the generator variables (h0_lin, biases, etc.) have None for gradients. I'm not sure why, or if that will change.
Now it is easy to change the defaults.
|
@MustafaMustafa , I wanted to ping you to see when you'll have time to look at this. Thanks. |
|
@MustafaMustafa - This should be much easier to look at. |
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
I'd like to use cosmoGAN as a simple GAN benchmark tool for compute nodes with multiple GPUs.
I've implemented the basic multiGPU support, but the discriminator loss doesn't converge.
I don't know why, and before I muck with it more I thought I'd ask for help and give back some of my modifications.
The modifications that work make it easy to override the default arguments on the command line.
There is a new option to display progress to stdout instead of a file.