def is_benchmark_supported()

in llm_perf/benchmark_runners/cuda/update_llm_perf_cuda_pytorch.py [0:0]

• 7 lines of code
• 3 McCabe index (conditional complexity)

    def is_benchmark_supported(self, **kwargs) -> bool:
        if (
            kwargs["attn_implementation"] == "flash_attention_2"
            and kwargs["weights_config"] == "float32"
        ):
            return False
        return True