[CI] Use a Stable Cache Key to prevent warnings in gh-actions

[ParagEkbote]

Description

In some of the gh-action test runs, I have observed the following warnings for the model and dataset cache not being saved correctly:

Warning: Failed to restore: The specified blob does not exist.
RequestId:b413aa80-601e-00a4-0531-699e57000000
Time:2025-12-09T17:30:01.1909781Z
Cache not found for input keys: hf-datasets-Linux-1bbd1d207191e929474a58aa111fdace73dbdc6ecb8d25a54146afc370d5d037, hf-datasets-Linux-

The issue stems from setting the cache key in an uncertain manner. When tests get updated or modified, a new key will be assigned and the previous cache will get invalidated.

In the updated key, we use a stable prefix hf-cache-${{ runner.os }}-, which cleanly separates caches by operating system. The addition of ${{ hashFiles('pyproject.toml') }} ensures the cache is only invalidated when your dependency configuration changes. This approach also enables partial key matching through restore keys, allowing gh-actions to fall back to compatible caches when an exact match can't be found.

Could you please review?

cc: @davidberenstein1957

Related Issue

Fixes #(issue number)

Type of Change

• Bug fix (non-breaking change which fixes an issue)
• New feature (non-breaking change which adds functionality)
• Breaking change (fix or feature that would cause existing functionality to not work as expected)
• This change requires a documentation update

How Has This Been Tested?

Checklist

• My code follows the style guidelines of this project
• I have performed a self-review of my own code
• I have commented my code, particularly in hard-to-understand areas
• I have made corresponding changes to the documentation
• I have added tests that prove my fix is effective or that my feature works
• New and existing unit tests pass locally with my changes

Additional Notes