[WIP] Update metrics.py - fix for ogbg pytorch #871
+7
−1
Conversation
This file contains hidden or 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
It seems that the problem affecting the pytorch ogbg workloads (but only if they run for some length of time) has to do with jax/xla cpu compilation of the metrics computation. By converting the jax arrays to numpy, hopefully this can be avoided. The next step is to test on schedule free and shampoo, which I hope to do very soon.
|
MLCommons CLA bot All contributors have signed the MLCommons CLA ✍️ ✅ |
|
You can see here that schedule free now completes the run with the changes. |
|
Tested again but this time only replacing the call to jax for sigmoid with numpy sigmoid. This also avoids the crash. So the problem seems to be calling jax sigmoid in pytorch. |
The problem with torchrun and jax seems to be caused by jax.nn.sigmoid.
Changed from lambda expression which pylint doesn't like.
Defined np sigmoid inside use_pytorch_ddp
Added white space before and after sigmoid_np
Fix white space
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Outdated suggestions cannot be applied.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.
It seems that the problem affecting the pytorch ogbg workloads (but only if they run for some length of time) has to do with jax/xla cpu compilation of the metrics computation. By converting the jax arrays to numpy, hopefully this can be avoided. The next step is to test on schedule free and shampoo, which I hope to do very soon.