We propose a state-of-the-art neural-based method for solving Time Dependent Traveling Salesman Problem and identify the limitation of existing evaluation method..