PyTorch and Torchvision version update

[tanwarsh]

Update PyTorch to version 2.4.1 and Torchvision to version 0.19.1.

Fix for #1075.

[tanwarsh]

Pipelines are facing Issue pytorch/pytorch#140914.

[kta-intel]

Pipelines are facing Issue pytorch/pytorch#140914.

Thanks for raising the issue. Can you look into the possibility of adding the monkey patch to openfl.interface.cli?
I didn't do any extensive testing on the e2e functionality, but it got past the TypeError

pytorch/pytorch#140914 (comment)

If it works, we can maybe implement it while we wait for a more robust upstream solution, or we can investigate where/how pip._vendor.typing_extensions is being forcibly called for a more robust solution in our code base

[tanwarsh]

Pipelines are facing Issue pytorch/pytorch#140914.

Thanks for raising the issue. Can you look into the possibility of adding the monkey patch to openfl.interface.cli? I didn't do any extensive testing on the e2e functionality, but it got past the TypeError

pytorch/pytorch#140914 (comment)

If it works, we can maybe implement it while we wait for a more robust upstream solution, or we can investigate where/how pip._vendor.typing_extensions is being forcibly called for a more robust solution in our code base

Thank you @kta-intel. Workaround patch worked.

[kta-intel]

@MasterSkepticista @kta-intel, We are using version 2.4.1 now without patch. Please review again.

Thanks for checking into that @tanwarsh. Glad 2.4.1 is not raising the TypeError issue. My open now is if downgrading setuptools is the path we want to go. Even if we're able to justify it for a release, if we're stuck at 59.6.0 going forward with torch workloads, that could be unsustainable long term

[tanwarsh]

@MasterSkepticista @kta-intel, We are using version 2.4.1 now without patch. Please review again.

Thanks for checking into that @tanwarsh. Glad 2.4.1 is not raising the TypeError issue. My open now is if downgrading setuptools is the path we want to go. Even if we're able to justify it for a release, if we're stuck at 59.6.0 going forward with torch workloads, that could be unsustainable long term

@kta-intel , I tried few things, trying different version and patches suggested in few issues logged in setuptools github repo but none of them are working. Looks like we are stuck with this solution.

[tanwarsh]

@psfoley @kta-intel @teoparvanov @rahulga1,
There are two solutions for the issue with Setuptools:

1. Pin Setuptools to version 59.6.0, which has a vulnerability issue. EXCEPTION : /python3.11/distutils/core.py pytorch/pytorch#143288.
2. Apply a patch to use the latest version of Setuptools (current changes).

Please review this PR and share your thoughts so that we can move forward.

[kta-intel]

@psfoley @kta-intel @teoparvanov @rahulga1, There are two solutions for the issue with Setuptools:

1. Pin Setuptools to version 59.6.0, which has a vulnerability issue. EXCEPTION : /python3.11/distutils/core.py pytorch/pytorch#143288.
2. Apply a patch to use the latest version of Setuptools (current changes).

Please review this PR and share your thoughts so that we can move forward.

I am okay with the patch for now, though we should keep it in the backlog to look for a more robust solution
Btw, do you know if this will break older versions (i.e. torch=2.3)?

[tanwarsh]

@psfoley @kta-intel @teoparvanov @rahulga1, There are two solutions for the issue with Setuptools:

1. Pin Setuptools to version 59.6.0, which has a vulnerability issue. EXCEPTION : /python3.11/distutils/core.py pytorch/pytorch#143288.
2. Apply a patch to use the latest version of Setuptools (current changes).

Please review this PR and share your thoughts so that we can move forward.

I am okay with the patch for now, though we should keep it in the backlog to look for a more robust solution Btw, do you know if this will break older versions (i.e. torch=2.3)?

This issue is not seen in torch==2.3.1, but the PR is to update torch from 2.3.1 because we were facing this issue #1075

[kta-intel]

@psfoley @kta-intel @teoparvanov @rahulga1, There are two solutions for the issue with Setuptools:

1. Pin Setuptools to version 59.6.0, which has a vulnerability issue. EXCEPTION : /python3.11/distutils/core.py pytorch/pytorch#143288.
2. Apply a patch to use the latest version of Setuptools (current changes).

Please review this PR and share your thoughts so that we can move forward.

I am okay with the patch for now, though we should keep it in the backlog to look for a more robust solution Btw, do you know if this will break older versions (i.e. torch=2.3)?

This issue is not seen in torch==2.3.1, but the PR is to update torch from 2.3.1 because we were facing this issue #1075

I was asking if patching os.environ["SETUPTOOLS_USE_DISTUTILS"] = "stdlib" would affect torch==2.3.1. If so, we should make note of it in a "known issues" section that a user would have to patch to local in their own application if they cannot upgrade, since we cannot expect a user to modify framework-level code, such as openfl/federated/__init__.py. But it doesn't, then it is no problem

[tanwarsh]

@psfoley @kta-intel @teoparvanov @rahulga1, There are two solutions for the issue with Setuptools:

1. Pin Setuptools to version 59.6.0, which has a vulnerability issue. EXCEPTION : /python3.11/distutils/core.py pytorch/pytorch#143288.
2. Apply a patch to use the latest version of Setuptools (current changes).

Please review this PR and share your thoughts so that we can move forward.

I am okay with the patch for now, though we should keep it in the backlog to look for a more robust solution Btw, do you know if this will break older versions (i.e. torch=2.3)?

This issue is not seen in torch==2.3.1, but the PR is to update torch from 2.3.1 because we were facing this issue #1075

I was asking if patching os.environ["SETUPTOOLS_USE_DISTUTILS"] = "stdlib" would affect torch==2.3.1. If so, we should make note of it in a "known issues" section that a user would have to patch to local in their own application if they cannot upgrade, since we cannot expect a user to modify framework-level code, such as openfl/federated/__init__.py. But it doesn't, then it is no problem

@kta-intel , sorry for the misunderstanding. I tested this in local. This patch works with version 2.3.1 as well. User will not have to modify anything.

[teoparvanov]

Hey @tanwarsh, thanks for upgrading the PyTorch workspaces to 2.4.1! I see this PR as an intermediate step on our "journey" to 2.5.