Abstract: Large Language Models (LLMs) have shown significant potential in automating software engineering tasks, particularly in code generation. However, current evaluation benchmarks, which ...