How to allocate more RAM and CPU in a pipeline?

In the following example, flip_coin is a component object.

@dsl.pipeline
def my_pipeline():
    coin_flip_task = flip_coin() \
        .set_cpu_limit(8) \
        .set_memory_limit("16G") \
        .add_node_selector_constraint("NVIDIA_TESLA_K80") \
        .set_gpu_limit(2)

Reference: https://cloud.google.com/vertex-ai/docs/pipelines/machine-types