[Auto Parallel]: Support optional, inplace input and output for DistTensor. #57092
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
Concat forward and backward are verified.
… dist_tensor_support
… dist_tensor_support
…Tensor> -> std::vector<Tensor>.
… dist_tensor_support
…or output of operators. Following testcases are passed. 1. concat: std::vector<phi::Tensor> -> phi::Tensor 2. unbind: phi::Tensor -> std::vector<phi::Tensor> 3. broadcast_tensors: std::vector<phi::Tensor> -> std::vector<phi::Tensor>
… dist_tensor_support
paddle::optional<std::vector<phi::Tensor>> input and output for DistTensor.
|
你的PR提交成功，感谢你对开源项目的贡献! |
GhostScreaming
changed the title
GhostScreaming
changed the title
… dist_tensor_support
chenwhql
reviewed
Sep 8, 2023
| OPTIONAL_GLOBAL_VECTOR_META_IN_TEMPLATE = """ | ||
| OPTIONAL_GLOBAL_SINGLE_META_IN_TEMPLATE = """meta_dist_{}, """ | ||
| OPTIONAL_GLOBAL_SINGLE_META_IN_DECL_TEMPLATE = """ | ||
| phi::MetaTensor meta_dist_{name} = {name} ? MakeMetaTensor(*(*{name}).impl()) : phi::MetaTensor(); |
There was a problem hiding this comment.
其实我们现在做了两遍meta tensor的构造，前面MakeDistMetaTensor也构造了一遍，后面我们可以看下能否在下面复用前面MakeDistMetaTensor的结果，从而降低调度开销
| 1, # align_mode | ||
| ) | ||
| # print("local_out: ", local_out) | ||
| dist_out = paddle._C_ops.linear_interp( |
There was a problem hiding this comment.
这个用正式api paddle.nn.functional.interpolate 可以测试吗
There was a problem hiding this comment.
需要测试linear_interp的paddle::optional<phi::Tensor> out_size和paddle::optional<std::vector<phi::Tensor>> size_tensor类型的输入，看了一下paddle.nn.functional.interpolate的代码实现，size_tensor是传不进去的。
…:optional<Tensor> and paddle::optional<std::vector<Tensor>>
GhostScreaming
changed the title
LiYuRio
reviewed
Sep 12, 2023
LiYuRio
reviewed
Sep 12, 2023
danleifeng
pushed a commit
to danleifeng/Paddle
that referenced
this pull request
Nov 14, 2023
…ensor. (PaddlePaddle#57092) * [WIP] Support std::vector<phi::Tensor> input and output for DistTensor. Concat forward and backward are verified. * Polish code for new dist tensor implementation. * Fix bug of DistTensor upgrade. Add support functions for std::vector<Tensor> -> std::vector<Tensor>. * Add support for DistTensor type of std::vector<phi::Tensor> as input or output of operators. Following testcases are passed. 1. concat: std::vector<phi::Tensor> -> phi::Tensor 2. unbind: phi::Tensor -> std::vector<phi::Tensor> 3. broadcast_tensors: std::vector<phi::Tensor> -> std::vector<phi::Tensor> * Polish code. Remove useless comments. * Add update_loss_scaling in skip_op_lists. * Polish code. * [Auto Parallel]: Support paddle::optional<Tensor> and paddle::optional<std::vector<phi::Tensor>> input and output for DistTensor. * Polish code. * Polish code. And support inplace Tensor, std::vector<Tensor>, paddle::optional<Tensor> and paddle::optional<std::vector<Tensor>> * Polish testcase code. Add testcase for inplace paddle::optional<phi::Tensor>. * Remove useless codes in testcase code. * Polish code style. * Polish code style. And fix problems of testcases.
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.
PR types
Others
PR changes
Others
Description
Pcard-73145
paddle::optional<phi::Tensor>andpaddle::optional<std::vector<phi::Tensor>>input and output forDistTensor.Expand_As,Bincount,Linear_Interpforward and backward are verified.inplace phi::Tensor,inplace std::vector<phi::Tensor>,inplace paddle::optional<phi::Tensor>andinplace paddle::optional<std::vector<phi::Tensor>>type of outputs.check_finite_and_unscale_,adamax_andmerged_adam_are verified.std::tuple<...>are supported now as all type of output are supported.