Fixes PyTorch 2.12-related incompatibility in fengwu model

[peterdsharpe]

PhysicsNeMo Pull Request

Description

Update fengwu_output.pth binary file to fix backwards-incompatible changes with PyTorch 2.12. See #1648 for similar issues; this one has the same failure mode.

Previous error:

FAILED test/models/fengwu/test_fengwu.py::test_fengwu_forward[cpu] - AssertionError: assert False
 +  where False = <function validate_forward_accuracy at 0x7fa3bd12f560>(Fengwu(\n  (encoder_surface): EncoderLayer(\n    (patchembed2d): PatchEmbed2D(\n      (pad): ZeroPad2d((0, 0, 0, 0))\n    ...atchrecovery2d): PatchRecovery2D(\n      (conv): ConvTranspose2d(384, 37, kernel_size=(4, 4), stride=(4, 4))\n    )\n  )\n), (tensor([[[[-8.7801e-01,  6.3541e-02,  3.9758e-01,  ...,  4.0209e-01,\n            6.2170e-01,  5.6378e-01],\n          ...1e+00],\n          [ 6.9760e-01,  2.6290e-01, -1.6820e-01,  ...,  4.2989e-01,\n           -1.6377e-01,  5.3725e-01]]]]),), atol=0.005, file_name='models/fengwu/data/fengwu_output.pth')
 +    where <function validate_forward_accuracy at 0x7fa3bd12f560> = common.validate_forward_accuracy

Checklist

• I am familiar with the Contributing Guidelines.
• New or existing tests cover these changes.
• The documentation is up to date with these changes.
• The CHANGELOG.md is up to date with these changes.
• An issue is linked to this pull request.
• If I am implementing a new model or modifying any existing model, I have followed the Models Implementation Coding Standards.

Dependencies

Review Process

All PRs are reviewed by the PhysicsNeMo team before merging.

Depending on which files are changed, GitHub may automatically assign a maintainer for review.

We are also testing AI-based code review tools (e.g., Greptile), which may add automated comments with a confidence score.
This score reflects the AI’s assessment of merge readiness and is not a qualitative judgment of your work, nor is
it an indication that the PR will be accepted / rejected.

AI-generated feedback should be reviewed critically for usefulness.
You are not required to respond to every AI comment, but they are intended to help both authors and reviewers.
Please react to Greptile comments with 👍 or 👎 to provide feedback on their accuracy.

[copy-pr-bot]

This pull request requires additional validation before any workflows can run on NVIDIA's runners.

Pull request vetters can view their responsibilities here.

Contributors can view more details about this message here.

[peterdsharpe]

/blossom-ci

[greptile-apps]

Greptile Summary

This PR updates the binary reference output file fengwu_output.pth used by test_fengwu_forward to reflect numerical differences introduced by PyTorch 2.12's backwards-incompatible changes, mirroring a fix already applied for similar models in #1648.

• The .pth file is a serialized tensor used as the ground-truth comparison in validate_forward_accuracy (with atol=5e-3). The file size is identical before and after, confirming tensor shapes are unchanged — only the numerical values differ due to PyTorch internals.
• No model code, test logic, or tolerance thresholds are modified; the fix is entirely contained in the reference data file.

Important Files Changed

Filename	Overview
test/models/fengwu/data/fengwu_output.pth	Binary reference output file regenerated for PyTorch 2.12 compatibility; file size unchanged (1551261 bytes), indicating tensor shapes are preserved.

_{Reviews (1): Last reviewed commit: "Update fengwu_output.pth binary file" | Re-trigger Greptile}

[ktangsali]

LGTM

[peterdsharpe]

/blossom-ci

[peterdsharpe]

/blossom-ci

[peterdsharpe]

/ok to test d62add6