[llama4] add apply_compile for moe, where fullgraph=False for moe layers #1519
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
meta-cla
bot
added
the
CLA Signed
xmfan
approved these changes
Aug 4, 2025
|
@tianyu-l i think fullgraph=False is better than compile just not working on main, so good with landing this |
tianyu-l
approved these changes
Aug 4, 2025
bentherien
pushed a commit
to bentherien/torchtitan_
that referenced
this pull request
Aug 5, 2025
…ers (pytorch#1519) We should add an `apply_compile` function for llama4 that uses fullgraph=False for MoE layers and fullgraph=True for dense layers. I keep manually applying this hack during development to test compile composability, but IMO we should have this merged and update to use fullgraph=True everywhere once that is supported. cc @xmfan @tianyu-l any thoughts?
joellidin
pushed a commit
to tplr-ai/torchtitan
that referenced
this pull request
Aug 8, 2025
…ers (pytorch#1519) We should add an `apply_compile` function for llama4 that uses fullgraph=False for MoE layers and fullgraph=True for dense layers. I keep manually applying this hack during development to test compile composability, but IMO we should have this merged and update to use fullgraph=True everywhere once that is supported. cc @xmfan @tianyu-l any thoughts?
joellidin
pushed a commit
to tplr-ai/torchtitan
that referenced
this pull request
Aug 8, 2025
…ers (pytorch#1519) We should add an `apply_compile` function for llama4 that uses fullgraph=False for MoE layers and fullgraph=True for dense layers. I keep manually applying this hack during development to test compile composability, but IMO we should have this merged and update to use fullgraph=True everywhere once that is supported. cc @xmfan @tianyu-l any thoughts?
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.
We should add an
apply_compilefunction for llama4 that uses fullgraph=False for MoE layers and fullgraph=True for dense layers.I keep manually applying this hack during development to test compile composability, but IMO we should have this merged and update to use fullgraph=True everywhere once that is supported.
cc @xmfan @tianyu-l any thoughts?