Remove more uses of DimensionedTensorType
#23060
Closed
Conversation
This file contains 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
pytorchbot
added
oncall: jit
Krovatkin
force-pushed
the
krovatkin/profiling_dim2
branch
from
3e724eb
to
67bc807
Compare
Krovatkin
changed the title
DimensionedTensorTypeDimensionedTensorType
Krovatkin
force-pushed
the
krovatkin/profiling_dim2
branch
from
070a914
to
0abb49f
Compare
Krovatkin
force-pushed
the
krovatkin/profiling_dim2
branch
from
f934329
to
fcc382e
Compare
There was a problem hiding this comment.
@Krovatkin has imported this pull request. If you are a Facebook employee, you can view this diff on Phabricator.
zdevito
approved these changes
Jul 29, 2019
| auto tensor_type = input->type()->cast<DimensionedTensorType>(); | ||
| // As of now, we do the decomposition for batchnorm/layernorm on GPU device only | ||
| if (!tensor_type || tensor_type->device().is_cpu()) { | ||
| if (!input->type()->isSubclass(TypeKind::TensorType)) { |
There was a problem hiding this comment.
The canonical way to do this is: !input->type()->isSubtype(TensorType::get())
| @@ -182,13 +182,16 @@ struct GraphFuser { | |||
| } | |||
|
|
|||
| bool isFusableDevice(Value *v) { | |||
| auto tensor_type = v->type()->cast<DimensionedTensorType>(); | |||
| if (!tensor_type) { | |||
| if (!v->type()->isSubclass(TypeKind::TensorType)) { | |||
| @@ -29,8 +29,10 @@ Node* InsertCastForCond(Value* cond_val, Graph* graph, Node* consumer_node) { | |||
|
|
|||
| bool IsCondCastRequired(Value* cond_val) { | |||
| const auto& type = cond_val->type(); | |||
| if (type->isSubclass(TypeKind::DimensionedTensorType)) { | |||
| return type->expect<DimensionedTensorType>()->scalarType() != c10::kBool; | |||
| if (type->isSubclass(TypeKind::TensorType)) { | |||
torch/jit/__init__.py
Outdated
| @@ -35,6 +35,13 @@ | |||
| if sys.version_info[0] > 2: | |||
| import pathlib | |||
|
|
|||
| def _get_type_attr(type, aname): | |||
There was a problem hiding this comment.
I'd rather have the 'dim' and scalarType methods handle returning None rather than throw/catch an exception. It makes things hard to debug (catch throw now catches spurious things), and it can mask a real exception that is not the one that is expected.
fix a bug in an assertion remove extra new lines more changes to remove DimensionedTensorType insert a newline assert -> torch_check only use `ProfiledTensorCreate` on tensors and insert checks elsewhere add a missing import fixing a call to _get_type_attr pass a type instead of a value into _get_type_attr don't rely on a tensor type to be dimensioned address Zach's feedback
apaszke
reviewed
Aug 1, 2019
Krovatkin
force-pushed
the
krovatkin/profiling_dim2
branch
from
c6068b3
to
f770a74
Compare
I think, that's the plan :-) eventually |
There was a problem hiding this comment.
@Krovatkin has imported this pull request. If you are a Facebook employee, you can view this diff on Phabricator.
zdevito
pushed a commit
to zdevito/ATen
that referenced
this pull request
Aug 2, 2019
Summary: Pull Request resolved: pytorch/pytorch#23060 Differential Revision: D16460391 Pulled By: Krovatkin fbshipit-source-id: b50ee87d22ad18b8cbfff719b199ea876ef172f1
|
@Krovatkin merged this pull request in 3d15ee1. |
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.
No description provided.