Fix: Handling empty sparsity config

[rahul-tuli]

FIX vllm-project#12044

This PR fixes a bug where the sparsity_config in the CompressedTensorsConfig class would not handle empty configurations correctly. The issue, tracked in vllm-project#12044, arose due to the explicit check for None values instead of evaluating the truthiness of sparsity_config. The fix ensures that empty configurations are now handled gracefully by directly checking the truthiness of sparsity_config.

Changes

• Updated the conditional check in CompressedTensorsConfig.get_sparsity_config to evaluate the truthiness of sparsity_config instead of explicitly comparing against None.

Testing

The fix was tested by running the following:

• nm-testing/llama2.c-stories42M-quantized-fp8-Dynamic
  • This test case includes an empty sparsity_config and confirmed the correctness of the fix.

Resolves

• Fixes [Bug]: ValidationError when loading fp8-dynamic model with empty "sparsity_config" vllm-project/vllm#12044

[github-actions]

👋 Hi! Thank you for contributing to the vLLM project.
Just a reminder: PRs would not trigger full CI run by default. Instead, it would only run fastcheck CI which starts running only a small and essential subset of CI tests to quickly catch errors. You can run other CI tests on top of those by going to your fastcheck build on Buildkite UI (linked in the PR checks section) and unblock them. If you do not have permission to unblock, ping simon-mo or khluu to add you in our Buildkite org.

Once the PR is approved and ready to go, your PR reviewer(s) can run CI to test the changes comprehensively before merging.

To run CI, PR reviewers can do one of these:

• Add ready label to the PR
• Enable auto-merge.

🚀