Add a contrib operator for dynamic reshape

[vandanavk]

Description

MXNet's reshape operator expects the shape to be a static tuple. In some models, the shape attribute is decided based on the output of the previous layer. Example, #10789. Adding support for allowing a tensor for the target shape in Reshape.

Based on #14393

Checklist

Essentials

Please feel free to remove inapplicable items for your PR.

• The PR title starts with [MXNET-$JIRA_ID], where $JIRA_ID refers to the relevant JIRA issue created (except PRs with tiny changes)
• Changes are complete (i.e. I finished coding on this PR)
• All changes have test coverage:
• Unit tests are added for small changes to verify correctness (e.g. adding a new operator)
• Nightly tests are added for complicated/long-running ones (e.g. changing distributed kvstore)
• Build tests will be added for build configuration changes (e.g. adding a new build option with NCCL)
• Code is well-documented:
• For user-facing API changes, API doc string has been updated.
• For new C++ functions in header files, their functionalities and arguments are documented.
• For new examples, README.md is added to explain the what the example does, the source of the dataset, expected performance on test set and reference to the original paper if applicable
• Check the API doc at http://mxnet-ci-doc.s3-accelerate.dualstack.amazonaws.com/PR-$PR_ID/$BUILD_ID/index.html
• To the my best knowledge, examples are either not affected by this change, or have been fixed to be compatible with this change

Changes

Dynamic reshape:

• Add a contrib operator for dynamic reshape
• Add test cases

Comments

Forward pass takes the same time as the original reshape operator

[vandanavk]

@mxnet-label-bot add [pr-work-in-progress, Operator, ONNX]

[Rafis1988]

Hi @vandanavk. I see that these changes have to fix onnx model import error. Sorry for a stupid question. How can I use them? Or I have to wait before it gets reviewed? I stuck on this problem for two weeks.

[vandanavk]

@mxnet-label-bot update [Operator, pr-awaiting-review]

[vandanavk]

@junrushao1994 for review

[vandanavk]

Hi @vandanavk. I see that these changes have to fix onnx model import error. Sorry for a stupid question. How can I use them? Or I have to wait before it gets reviewed? I stuck on this problem for two weeks.

You might have to pull in changes from all the following PRs into your copy of mxnet, build and then try to convert. The PRs are #15872, #15993, #15995, #15996

[apeforest]

@samskalicky Would you mind reviewing this PR since it's related to dynamic reshape?

[samskalicky]

Thanks for this contribution @vandanavk! Sorry its been sitting so long without review. Theres a few things i dont understand, i left some comments. could you please clarify?

[vandanavk]

Thanks for this contribution @vandanavk! Sorry its been sitting so long without review. Theres a few things i dont understand, i left some comments. could you please clarify?

thanks for the review @samskalicky. let me know if the responses make things a little more clear

[apeforest]

Thanks for your contribution. Left a few comments.

[vandanavk]

@samskalicky @apeforest Addressed all review comments so far

[apeforest]

Thanks! LGTM

[apeforest]

@zheng-da may be interested to take a look?