Multiple Choice

Why does Machine Learning Yearning consider the non-end-to-end approach more promising for autonomous driving?