Dichen contributed to the vllm-project/llm-compressor repository by expanding unit test coverage for the ModuleSparsificationInfo class, focusing on validating quantization and sparsity parameter calculations. Using Python, PyTorch, and Pytest, Dichen developed a comprehensive test suite targeting the params_quantized_percent and params_sparse_percent methods. This work aimed to strengthen the validation logic for quantization and sparsity, reducing the risk of regressions in model compression workflows. By improving test coverage, Dichen enhanced code quality and supported safer deployment of sparsification features. The depth of the work reflects a methodical approach to ensuring robust and maintainable model compression infrastructure.