Skip to content
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

This is a common operation, but ONNX doesn't give us a convenient way of expressing this #1539

Open
justinchuby opened this issue May 15, 2024 · 1 comment
Labels
topic: torch_lib Related to the torch/aten function lib in development

Comments

@justinchuby
Copy link
Contributor

          This is a common operation, but ONNX doesn't give us a convenient way of expressing this. We could potentially extend either the [Flatten op](https://onnx.ai/onnx/operators/onnx__Flatten.html#l-onnx-doc-flatten) or [Reshape op](https://onnx.ai/onnx/operators/onnx__Reshape.html) to support this. This is a one-sided flatten. Alternatively, if reshape has an attribute to align the shape with existing shape starting from the end (instead of beginning), we could specify the new target shape as `[-1, 0, 0, 0]`. Just some idle thoughts ...

Originally posted by @gramalingam in #1514 (comment)

@titaiwangms
Copy link
Contributor

Is this supposed to be changing ops in pixelshuffle? I think we should make it more clear to look actionable.

@titaiwangms titaiwangms added the topic: torch_lib Related to the torch/aten function lib in development label May 15, 2024
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
topic: torch_lib Related to the torch/aten function lib in development
Projects
None yet
Development

No branches or pull requests

2 participants