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

Use Attach TPC to Framework as part of MCT main track #1308

Merged
merged 44 commits into from
Jan 6, 2025

Conversation

ofirgo
Copy link
Collaborator

@ofirgo ofirgo commented Jan 1, 2025

Pull Request Description:

This PR changes the TPC mechanism in MCT APIs.
All the main API functions now get a framework-agnostic target platform description and use the Attach2Fw module to generate a framework-dependent quantization capabilities description.

The main changes include:

  • Alignment of the existing TP models with the new schema (v1).
  • Remove TP models v2-4 for IMX500 (moved to tests for simplicity of testing).
  • Remove existing fw TPCs (no need for them anymore).
  • Addition of attach2fw usage in all API functions.
    • To prevent breaking changes for users using the default TPC, the get_target_platform_capabilities API function is kept with a reduced functionality that allows to only obtain the default v1 TPC.
  • Minor changes to the built-in OperatorSetsNames enum to support more required operators.
  • Massive tests modification and alignment to comply with the new TPC mechanism.

Checklist before requesting a review:

  • I set the appropriate labels on the pull request.
  • I have added/updated the release note draft (if necessary).
  • I have updated the documentation to reflect my changes (if necessary).
  • All function and files are well documented.
  • All function and classes have type hints.
  • There is a licenses in all file.
  • The function and variable names are informative.
  • I have checked for code duplications.
  • I have added new unittest (if necessary).

ofirgo and others added 21 commits December 25, 2024 12:33
Replace schema classes from dataclass to pydantic 'BaseModel'.
Fix tests to support pydantic schema classes.
Add test for exporting tp model to json
All Pytorch Feature model tests pass except few special cases to solve later.
…al tpc and dedicated tests for the TPC versions
- enforce matching opsets from tp model to fw
- modify attach2fw classes names
- remove unused static tpc initialization
@ofirgo
Copy link
Collaborator Author

ofirgo commented Jan 1, 2025

@lior-dikstein @irenaby
Please start going over the main code and API changes.
Note that the work on the tests is still not done, so don't go over the tests yet.
Thanks

@ofirgo ofirgo merged commit d895518 into sony:main Jan 6, 2025
40 checks passed
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Projects
None yet
Development

Successfully merging this pull request may close these issues.

4 participants