Learning to Generate Hard Instances: Towards Robust Solutions for Vehicle Routing Problems

Dependencies

• Python >= 3.10

• pytorch == 2.0.1

• pyCombinatorial (for attacking heuristics)

How to run

python .\run.py -train=True --env="POMO" --problem="TSP" --target_size="100"

Please use python run.py --help for more information about the configurations.

Acknowledgements

• GitHub - yd-kwon/POMO: codes for the paper "POMO: Policy Optimization with Multiple Optima for Reinforcement Learning"

• GitHub - wondergo2017/TSP-HAC: Codes for AAAI22 paper "Learning to Solve Travelling Salesman Problem with Hardness-Adaptive Curriculum"

• GitHub - yd-kwon/MatNet: codes for the paper "Matrix Encoding Networks for Neural Combinatorial Optimization"

• GitHub - RoyalSkye/Omni-VRP: [ICML 2023] "Towards Omni-generalizable Neural Methods for Vehicle Routing Problems"

• GitHub - jieyibi/AMDKD: The implementation code of our paper "Learning Generalizable Models for Vehicle Routing Problems via Knowledge Distillation", accepted at NeurIPS2022.

• GitHub - yd-kwon/SGBS: codes for the paper "Simulation-guided Beam Search for Neural Combinatorial Optimization"