[aoti] Fix constant inputs passed to aoti #131594
Closed
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
angelayi
requested review from
avikchaudhuri,
tugsbayasgalan,
ydwu4 and
zhxchen17
as code owners
🔗 Helpful Links🧪 See artifacts and rendered test results at hud.pytorch.org/pr/131594
Note: Links to docs will display an error until the docs builds have been completed. ✅ No FailuresAs of commit 16f315b with merge base 980bb54 ( This comment was automatically generated by Dr. CI and updates every 15 minutes. |
Contributor
|
@angelayi has imported this pull request. If you are a Meta employee, you can view this diff on Phabricator. |
angelayi
force-pushed
the
angelayi/skip_consts
branch
from
5a9b27a to
bab80fa
Compare
Contributor
|
@angelayi has imported this pull request. If you are a Meta employee, you can view this diff on Phabricator. |
angelayi
force-pushed
the
angelayi/skip_consts
branch
from
bab80fa to
22cf510
Compare
desertfire
approved these changes
Jul 24, 2024
| in_spec = pytree.treespec_loads(call_spec[0]) | ||
| out_spec = pytree.treespec_loads(call_spec[1]) | ||
| flat_inputs = pytree.tree_flatten((args, reorder_kwargs(kwargs, in_spec)))[0] | ||
| flat_inputs = [x for x in flat_inputs if isinstance(x, torch.Tensor)] |
Contributor
There was a problem hiding this comment.
What about the C++ side? Is there a way to extend the Runner interface?
Contributor
Author
There was a problem hiding this comment.
Hmm, right now it results in a type error if you pass non-tensors to the runner because the cpp interface is for a List[Tensor]. Does this mean that we should convert it to being a union of inputs?
angelayi
force-pushed
the
angelayi/skip_consts
branch
from
22cf510 to
16f315b
Compare
Contributor
|
@angelayi has imported this pull request. If you are a Meta employee, you can view this diff on Phabricator. |
Contributor
|
@pytorchbot merge -f 'Landed internally' (Initiating merge automatically since Phabricator Diff has merged, using force because this PR might not pass merge_rules.json but landed internally) |
Collaborator
Merge startedYour change will be merged immediately since you used the force (-f) flag, bypassing any CI checks (ETA: 1-5 minutes). Please use Learn more about merging in the wiki. Questions? Feedback? Please reach out to the PyTorch DevX Team |
This was referenced Aug 7, 2024
coconutruben
added a commit
to coconutruben/pytorch
that referenced
this pull request
Sep 30, 2024
Summary: # Why The arguments are filtered out as they are just const in the compiled graph, but the logger still expects a non-None type # What When passing a filtered out arg (None) to the debug logger, just log that it's a filtered out argument, instead of throwing a Type error # Background pytorch#131594 Test Plan: - execute repro from pytorch#135584 (comment) with and without the edits Differential Revision: D63652564
coconutruben
added a commit
to coconutruben/pytorch
that referenced
this pull request
Sep 30, 2024
…#137032) Summary: Pull Request resolved: pytorch#137032 # Why The arguments are filtered out as they are just const in the compiled graph, but the logger still expects a non-None type # What When passing a filtered out arg (None) to the debug logger, just log that it's a filtered out argument, instead of throwing a Type error also add unit test to capture this behavior going forward # Background pytorch#131594 Test Plan: - execute repro from pytorch#135584 (comment) with and without the edits Reviewed By: angelayi Differential Revision: D63652564
coconutruben
added a commit
to coconutruben/pytorch
that referenced
this pull request
Oct 1, 2024
…#137032) Summary: Pull Request resolved: pytorch#137032 # Why The arguments are filtered out as they are just const in the compiled graph, but the logger still expects a non-None type # What When passing a filtered out arg (None) to the debug logger, just log that it's a filtered out argument, instead of throwing a Type error also add unit test to capture this behavior going forward # Background pytorch#131594 Test Plan: - execute repro from pytorch#135584 (comment) with and without the edits Reviewed By: angelayi Differential Revision: D63652564
coconutruben
added a commit
to coconutruben/pytorch
that referenced
this pull request
Oct 1, 2024
…#137032) Summary: Pull Request resolved: pytorch#137032 # Why The arguments are filtered out as they are just const in the compiled graph, but the logger still expects a non-None type # What When passing a filtered out arg (None) to the debug logger, just log that it's a filtered out argument, instead of throwing a Type error also add unit test to capture this behavior going forward # Background pytorch#131594 Test Plan: - execute repro from pytorch#135584 (comment) with and without the edits Reviewed By: angelayi Differential Revision: D63652564
pytorchmergebot
pushed a commit
that referenced
this pull request
Oct 2, 2024
Summary: # Why The arguments are filtered out as they are just const in the compiled graph, but the logger still expects a non-None type # What When passing a filtered out arg (None) to the debug logger, just log that it's a filtered out argument, instead of throwing a Type error # Background #131594 Test Plan: - execute repro from #135584 (comment) with and without the edits Differential Revision: D63652564 Pull Request resolved: #137032 Approved by: https://github.com/angelayi
AnantGulati
pushed a commit
to AnantGulati/pytorch
that referenced
this pull request
Oct 2, 2024
…#137032) Summary: # Why The arguments are filtered out as they are just const in the compiled graph, but the logger still expects a non-None type # What When passing a filtered out arg (None) to the debug logger, just log that it's a filtered out argument, instead of throwing a Type error # Background pytorch#131594 Test Plan: - execute repro from pytorch#135584 (comment) with and without the edits Differential Revision: D63652564 Pull Request resolved: pytorch#137032 Approved by: https://github.com/angelayi
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.
In cases where the program takes in a constant, export will specialize on the constant and embed the constant into the graph, with the graph containing a placeholder node with no users. However, inductor errors further down as typically in torch.compile, these constants don't show up as inputs. Since these constants are already embedded in the graph, we will just ignore these inputs while compiling with AOTI, and filter out the non-tensor inputs during the runtime.
cc @voznesenskym @penguinwu @EikanWang @jgong5 @Guobing-Chen @XiaobingSuper @zhuhaozhe @blzheng @wenzhe-nrv @jiayisunx @ipiszy @yf225 @chenyang78 @kadeng @muchulee8 @ColinPeppler @amjames @desertfire @chauhang