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

[MO] Relax Reshape layer hardcode under MatMul #1921

Merged

Conversation

jane-intel
Copy link
Contributor

Description: Transformation to relax hard-coded Reshape to 2D before MatMul operation

JIRA: as a part of #37084

Code:

  • Comments
  • Code style (PEP8)
  • Transformation generates reshape-able IR EVEN BETTER :) IT MAKES IR reshape-able
  • Transformation preserves original framework node names

Validation:

  • Unit tests
  • Framework operation tests
  • Transformation tests
  • e2e model test with an update of ./tests/e2e_oss/utils/reshape_utils.py
  • Model Optimizer IR Reader check

Documentation:

  • Supported frameworks operations list
  • Supported public models list
  • New operations specification
  • Guide on how to convert the public model
  • User guide update

@jane-intel jane-intel added the category: MO Model Optimizer label Aug 24, 2020
@jane-intel jane-intel added this to the 2021.1 milestone Aug 24, 2020
@jane-intel jane-intel requested review from lazarevevgeny, yekruglov and a team August 24, 2020 16:31
@vurusovs
Copy link
Contributor

@jane-intel please, take a look at memory tests - there is some significant increase of consumption for VGG16 on CPU and GPU (you may observe timeline report in "stress" job to compare with previous commits' validation)

@jane-intel
Copy link
Contributor Author

@jane-intel please, take a look at memory tests - there is some significant increase of consumption for VGG16 on CPU and GPU (you may observe timeline report in "stress" job to compare with previous commits' validation)

I do not see any fails in the stress job. Also there is no place from where some significant increase of memory consumption can come.

@jane-intel
Copy link
Contributor Author

@lazarevevgeny, there was and still is and intention to make it ND pass where possible. 2D case is the most popular, so I would like to merge it right now and work on extending the pass in the next release.

@lazarevevgeny lazarevevgeny merged commit 4507096 into openvinotoolkit:master Aug 27, 2020
RomanZm pushed a commit to RomanZm/openvino that referenced this pull request Aug 28, 2020
* [MO] Relax Reshape layer hardcode under MatMul

* Memory fix
@jane-intel jane-intel deleted the mo/evgenya/smart-reshape branch May 26, 2023 08:48
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
category: MO Model Optimizer
Projects
None yet
Development

Successfully merging this pull request may close these issues.

3 participants