-
Notifications
You must be signed in to change notification settings - Fork 576
pin update #7273
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
pin update #7273
Conversation
🔗 Helpful Links🧪 See artifacts and rendered test results at hud.pytorch.org/pr/pytorch/executorch/7273
Note: Links to docs will display an error until the docs builds have been completed. ✅ No FailuresAs of commit 64635c1 with merge base 18142f7 ( This comment was automatically generated by Dr. CI and updates every 15 minutes. |
13336ac
to
eb07fb2
Compare
Fix related to training IR migration for arm unittest: |
7798e48
to
e14318f
Compare
QNN failure potentially due to functionalization changes: executorch/extension/llm/export/builder.py Lines 187 to 194 in 46acc71
|
e14318f
to
656ac39
Compare
extension/llm/export/builder.py
Outdated
@@ -193,6 +193,7 @@ def export(self) -> "LLMEdgeManager": | |||
dynamic_shapes=dynamic_shape, | |||
strict=True, | |||
) | |||
exported_module.run_decompositions({}) |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
You need to re-assign exported_module, it is not an inplace update function.
7ae005e
to
844fa60
Compare
fix for arm test: pytorch/pytorch#143376 |
f3845fe
to
a3d70a5
Compare
test-arm-backend-delegation failure appears to be a known problem fixed by @digantdesai in pytorch/pytorch#143376 14 hours ago; probably need to roll the version forward? (might also need to wait a day; not sure if it's hit a nightly yet or not) |
a3d70a5
to
7b8e6bd
Compare
Yeah - looks like it's in the 2024-12-18 nightly, updating: https://github.com/pytorch/pytorch/blob/2ea4b56ec872424e486c4fe2d55da061067a2ed3/torch/fx/passes/infra/partitioner.py#L184 |
f40a8b9
to
d3b0371
Compare
d3b0371
to
3451182
Compare
adding patch for qnn to use old export flow: #7370 |
83f7539
to
c3ee8f6
Compare
verbose=self.verbose, | ||
) | ||
|
||
override_export_behaviour = contextlib.nullcontext() |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Can you leave a TODO here and mention the issue you filed?
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
ah - I left it at the top of the block.
d2f95cb
to
ebe5d6a
Compare
Pytorch nightly now sets _GLIBCXX_USE_CXX11_ABI by default. Think this is causing the errors in llava runner. |
b8256dc
to
c53b550
Compare
c53b550
to
64635c1
Compare
@@ -298,7 +298,6 @@ prebuilt_cxx_library( | |||
name = "libtorch", | |||
shared_lib = ":libtorch_gen[libtorch]", | |||
exported_preprocessor_flags = [ | |||
"-D_GLIBCXX_USE_CXX11_ABI=0", # `libtorch` is built without CXX11_ABI so any target depends on it need to use the same build config. |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Let's see if this works. @atalman Do you know if we have switch libtorch build to CXX11_ABI yet? I don't see it in pytorch/pytorch#143423 yet
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
All the tests are green! Good to land? @huydhn
Rebasing #7235
Changes: https://github.com/pytorch/pytorch/compare/19eff28ff3f19b50da46f5a9ff5f4d4d213806fe..80ca6dd892613fd4f1dee9040b8273ddeadb1c50