Add dry run mode #2012
Merged
Add dry run mode #2012
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
[ghstack-poisoned]
fegin
added a commit
that referenced
this pull request
Nov 10, 2025
Summary: The current configuration validation requires torchx and GPUs. It can waste time, resources, ane engery. Polar bears are crying. Let's fix this by providing a dry run mode. This PR doesn't verify everything. In theory, we should be able to verify parallelisms settings as well. This PR is just a start but it at least can let us catch the typos quickly. ghstack-source-id: 8ffb03d Pull-Request: #2012
meta-cla
bot
added
the
CLA Signed
tianyu-l
reviewed
Nov 10, 2025
Contributor
tianyu-l
left a comment
tianyu-l
left a comment
There was a problem hiding this comment.
Wish we could keep train.py and run_train.sh simple..
Contributor
Author
|
Okay, there is one minor issue by have a separate DryRunTrainer. More and more applications are subclassing Trainer. So if we create another DryRunTrainer, then we don't benefit those applications (e.g., full dtensor). So I guess we will eventually merge this back once we figure out how to do the full dry run with fake backend (after the DeviceMesh PR is landed). |
fegin
added a commit
that referenced
this pull request
Nov 11, 2025
Summary: The current configuration validation requires torchx and GPUs. It can waste time, resources, ane engery. Polar bears are crying. Let's fix this by providing a dry run mode. This PR doesn't verify everything. In theory, we should be able to verify parallelisms settings as well. This PR is just a start but it at least can let us catch the typos quickly. ghstack-source-id: 72f4316 Pull-Request: #2012
[ghstack-poisoned]
fegin
added a commit
that referenced
this pull request
Nov 11, 2025
Summary: The current configuration validation requires torchx and GPUs. It can waste time, resources, ane engery. Polar bears are crying. Let's fix this by providing a dry run mode. This PR doesn't verify everything. In theory, we should be able to verify parallelisms settings as well. This PR is just a start but it at least can let us catch the typos quickly. ghstack-source-id: 4e599a7 Pull-Request: #2012
tianyu-l
approved these changes
Nov 11, 2025
Contributor
tianyu-l
left a comment
tianyu-l
left a comment
There was a problem hiding this comment.
Had a suggestion on where to put the dry_run.py file.
Agree that we should switch to fake backend if that works.
Also maybe LocalTensor can be helpful?
Contributor
There was a problem hiding this comment.
nit: Would you consider putting this under torchtitan/tools/dry_run.py (or other 2nd level directory under torchtitan), or scripts/dry_run.py?
Contributor
Author
There was a problem hiding this comment.
Okay, let's put it under scripts/dry_run.py for now. We should investigate how to merge it back to train.py with LocalTensor or fake backend anyway.
[ghstack-poisoned]
fegin
added a commit
that referenced
this pull request
Nov 11, 2025
Summary: The current configuration validation requires torchx and GPUs. It can waste time, resources, ane engery. Polar bears are crying. Let's fix this by providing a dry run mode. This PR doesn't verify everything. In theory, we should be able to verify parallelisms settings as well. This PR is just a start but it at least can let us catch the typos quickly. ghstack-source-id: 99e031f Pull-Request: #2012
[ghstack-poisoned]
fegin
added a commit
that referenced
this pull request
Nov 11, 2025
Summary: The current configuration validation requires torchx and GPUs. It can waste time, resources, ane engery. Polar bears are crying. Let's fix this by providing a dry run mode. This PR doesn't verify everything. In theory, we should be able to verify parallelisms settings as well. This PR is just a start but it at least can let us catch the typos quickly. ghstack-source-id: 7c01387 Pull-Request: #2012
fegin
added a commit
that referenced
this pull request
Nov 11, 2025
Stack from [ghstack](https://github.com/ezyang/ghstack/tree/0.12.0) (oldest at bottom): * #2012 * __->__ #2011 It is not correct as JobConfig has changed.
ahoffman-aws
pushed a commit
to drcanchi-aws/torchtitan
that referenced
this pull request
Nov 11, 2025
Stack from [ghstack](https://github.com/ezyang/ghstack/tree/0.12.0) (oldest at bottom): * pytorch#2012 * __->__ pytorch#2011 It is not correct as JobConfig has changed.
ahoffman-aws
pushed a commit
to drcanchi-aws/torchtitan
that referenced
this pull request
Nov 11, 2025
Stack from [ghstack](https://github.com/ezyang/ghstack/tree/0.12.0) (oldest at bottom): * __->__ pytorch#2012 * pytorch#2011 Summary: The current configuration validation requires torchx and GPUs. It can waste time, resources, ane engery. Polar bears are crying. Let's fix this by providing a dry run mode. This PR doesn't verify everything. In theory, we should be able to verify parallelisms settings as well. This PR is just a start but it at least can let us catch the typos quickly.
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.
Stack from ghstack (oldest at bottom):
Summary:
The current configuration validation requires torchx and GPUs. It can waste time, resources, ane engery. Polar bears are crying. Let's fix this by providing a dry run mode. This PR doesn't verify everything. In theory, we should be able to verify parallelisms settings as well. This PR is just a start but it at least can let us catch the typos quickly.