Add tlparse support #21
Merged
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
facebook-github-bot
added
the
CLA Signed
bdhirsh
approved these changes
Jun 26, 2025
Contributor
bdhirsh
left a comment
bdhirsh
left a comment
There was a problem hiding this comment.
definitely looks to me like a good starting point!
Run your pytorch program with `TORCH_TRACE=<path_to_logdir>` set, and then run `tlparse <path_to_specific_logfile>` to generate a manifold URL to an html artifact. At this point, it adds 3 artifact files into the tlparse, and others could be added later as needed. - autoparallel_joint_graph - autoparallel_sharding_optimizer_log - autoparallel_parallel_graph Example output (llama3 debugmodel, 8gpu, tp=4, without torch.compile): https://manifold.edge.x2p.facebook.net/v0/read/tree/logs/.tmpblWM9X/index.html?bucketName=tlparse_reports&apiKey=tlparse_reports-key&withPayload=1&timeoutMsec=10000
fmassa
reviewed
Jun 27, 2025
| # clean it up by removing the added aliases from previous pass | ||
| # as well as redundant views | ||
| parallel_gm = joint_graph_passes(parallel_gm) | ||
| trace_structured( |
Contributor
There was a problem hiding this comment.
nit: it would be maybe more readable to store the parallel_gm after the apply_node_renaming as the nodes would have the same name as for the unsharded graph
fmassa
reviewed
Jun 27, 2025
| "name": "autoparallel_joint_graph", | ||
| "encoding": "string", | ||
| }, | ||
| payload_fn=lambda: str(gm.graph), |
Contributor
There was a problem hiding this comment.
Might be preferable to store the str(gm), as it would let it be a runnable representation of the graph.
bdhirsh
pushed a commit
that referenced
this pull request
Jul 1, 2025
Run your pytorch program with `TORCH_TRACE=<path_to_logdir>` set, and then run `tlparse <path_to_specific_logfile>` to generate a manifold URL to an html artifact. At this point, it adds 3 artifact files into the tlparse, and others could be added later as needed. - autoparallel_joint_graph - autoparallel_sharding_optimizer_log - autoparallel_parallel_graph Example output (llama3 debugmodel, 8gpu, tp=4, without torch.compile): https://manifold.edge.x2p.facebook.net/v0/read/tree/logs/.tmpblWM9X/index.html?bucketName=tlparse_reports&apiKey=tlparse_reports-key&withPayload=1&timeoutMsec=10000
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.
Run your pytorch program with
TORCH_TRACE=<path_to_logdir>set, and then runtlparse <path_to_specific_logfile>to generate a manifold URL to an html artifact.At this point, it adds 3 artifact files into the tlparse, and others could be added later as needed.
Example output (llama3 debugmodel, 8gpu, tp=4, without torch.compile): https://manifold.edge.x2p.facebook.net/v0/read/tree/logs/.tmpblWM9X/index.html?bucketName=tlparse_reports&apiKey=tlparse_reports-key&withPayload=1&timeoutMsec=10000