This repository was archived by the owner on Nov 15, 2022. It is now read-only.
Updated max_pool2d for same size tensors #157
Merged
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
izdeby
commented
May 29, 2020
| ceil_mode); | ||
|
|
||
| return at::detail::make_tensor<NestedTensorImpl>( | ||
| torch::nested_tensor::NestedTensor(std::move(res)).to_nested_tensor(nt.nested_dim() - 1)); |
There was a problem hiding this comment.
Im concerned about nt.nested_dim() - 1
There was a problem hiding this comment.
Throw an error if nested_dim() is below 1 (which it never should be)
cpuhrsch
reviewed
May 29, 2020
| for nt in [nestedtensor.nested_tensor(inputs), nestedtensor.as_nested_tensor(inputs)]: | ||
| nt_res = maxPool2d(nt) | ||
| self.assertEqual(nestedtensor.nested_tensor(tensor_res), nt_res) | ||
| for nt in [nestedtensor.nested_tensor(inputs), nestedtensor.as_nested_tensor(inputs)]: |
There was a problem hiding this comment.
You don't need to check as_nested_tensor anymore since it'll do the same thing as nested_tensor in this case (we changed that in #150).
cpuhrsch
reviewed
May 29, 2020
cpuhrsch
reviewed
May 29, 2020
cpuhrsch
reviewed
Jun 2, 2020
| auto nt = self_impl->_data; | ||
| auto tensor_node = get_nested_tensor_structure(self); | ||
|
|
||
| if (is_tensor_shape(self)) { |
There was a problem hiding this comment.
You can make this a method, but I can do that later on as part of further work
cpuhrsch
reviewed
Jun 2, 2020
| ceil_mode); | ||
|
|
||
| return at::detail::make_tensor<NestedTensorImpl>( | ||
| torch::nested_tensor::NestedTensor(std::move(res)).to_nested_tensor(nt.nested_dim() - 1)); |
There was a problem hiding this comment.
There's a helper for wrapping a C++ NestedTensor called wrap_nested_tensor
cpuhrsch
approved these changes
Jun 2, 2020
Sign up for free
to subscribe to this conversation on GitHub.
Already have an account?
Sign in.
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.
Benchmark setup:
C = 256
H/W = 64
Variance = 0
Benchmark results(CPU):
Master:
N = 8
('name', 'max_pool2d_tensor_pad'),('runs', 2867)
('name', 'max_pool2d_nt'),('runs', 362)
N = 16
('name', 'max_pool2d_tensor_pad'),('runs', 2222)
('name', 'max_pool2d_nt'),('runs', 180)
This change:
N = 8
('name', 'max_pool2d_tensor_pad'),('runs', 2869)
('name', 'max_pool2d_nt'),('runs', 2347)
N = 16
('name', 'max_pool2d_tensor_pad'),('runs', 2214)
('name', 'max_pool2d_nt'),('runs', 1639)