You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
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
1. Semantic of List/Tensor and Tuple: In indexing, the semantics of List type is same with Tensor/Array type (corresponding to one axis). As a comparison,Tuple type is different (corresponding to multiple axes, each element is an index on the corresponding axis).
2. The current semantic of List are contradictory:Currently in Paddle, List type have two semantics.
If rank of List is one, it corresponds to one axis (same with Tensor/Array, which is naturally).
if its rank is greater than 1, it means that it may correspond to multiple axes (The outermost [] is equivalent to tuple,which is to wrap indexes on different axes together).
# In this case, List is same with Tuple, and different with Tensor/Array>>>a[[[0],[1]]]
Tensor(shape=[1], dtype=float32, place=Place(gpu:0), stop_gradient=True,
[0.19835377])
>>>a[([0],[1])]
Tensor(shape=[1], dtype=float32, place=Place(gpu:0), stop_gradient=True,
[0.19835377])
>>>a[paddle.to_tensor([[0],[1]])]
Tensor(shape=[2, 1, 2], dtype=float32, place=Place(gpu:0), stop_gradient=True,
[[[ 0.07773950, 0.19835377]],
[[-0.81423134, 0.60788709]]])
What this PR did
This PR unified the semantics of List in indexing, like Numpy(as described in #51466 ). From this PR, List is same with Tensor/Array in any case.
你的PR提交成功,感谢你对开源项目的贡献!
请关注后续CI自动化测试结果,详情请参考Paddle-CI手册。
Your PR has been submitted. Thanks for your contribution!
Please wait for the result of CI firstly. See Paddle CI Manual for details.
The fundamental modification of thist PR is to unify the tensor indexing semantics of one-dimensional list and multidimensional list. So, it is recommended to use tensor indexing of one-dimensional list and tensor indexing of multidimensional list for comparison in the examples of description.
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.
PR types
Breaking changes
PR changes
APIs
Description
Pcard-66985
Background Information:
1. Semantic of List/Tensor and Tuple: In indexing, the semantics of
List
type is same withTensor/Array
type (corresponding to one axis). As a comparison,Tuple
type is different (corresponding to multiple axes, each element is an index on the corresponding axis).2. The current semantic of List are contradictory:Currently in Paddle,
List
type have two semantics.Tensor/Array
, which is naturally).[]
is equivalent to tuple,which is to wrap indexes on different axes together).What this PR did
This PR unified the semantics of
List
in indexing, like Numpy(as described in #51466 ). From this PR,List
is same withTensor/Array
in any case.