Function get_memory_consumption_stat

tests/quantization/utils.py:39–45

(model, inputs)

    @torch.no_grad()
    @torch.inference_mode()
    def get_memory_consumption_stat(model, inputs):
        backend_reset_peak_memory_stats(torch_device)
        backend_empty_cache(torch_device)

        model(**inputs)
        max_mem_allocated = backend_max_memory_allocated(torch_device)
        return max_mem_allocated