ADVERSARIAL GENERATIVE FLOWNETWORK FOR SOLVING VEHICLE ROUTING PROBLEMS

Adversarial Generative Flow Network for Solving Vehicle Routing Problems

Our paper is accepted by ICLR 2025. For the paper associated with this codebase, see the link above.

CVRP

Training

$cd cvrp
$python train_cvrp.py $N

$N is the nodes of the instances

Testing

$cd cvrp
$python test_cvrp.py $N -p "path_to_checkpoint"

TSP

Training

$cd tsp
$python train_tsp.py $N

$N is the nodes of the instances

Testing

$cd tsp
$python test_tsp.py $N -p "path_to_checkpoint"

Reference

If you find this codebase useful, please consider citing the paper:

@inproceedings{zhangadversarial,
  title={Adversarial Generative Flow Network for Solving Vehicle Routing Problems},
  author={Zhang, Ni and Yang, Jingfeng and Cao, Zhiguang and Chi, Xu},
  booktitle={The Thirteenth International Conference on Learning Representations},
  year={2025}
}