CaDA

The official code repository for CaDA: Cross-Problem Routing Solver with Constraint-Aware Dual-Attention.

Preparation

1. Clone the repository:

git clone https://github.com/CIAM-Group/CaDA.git

2. Download datasets:

• Download data.zip from Hugging Face.
• Unzip data.zip and organize the files in the project directory as follows:

CaDA
├── data
│   ├── lib_data
│   └── synthetic_data
├── 50
├── 100
└── utils

3. Download checkpoints:

• Create 'result' folder manually under 'CaDA/50' and 'CaDA/100'.
• Download checkpoint.zip from Hugging Face.
• Unzip checkpoint.zip. It will produce two directories: 50 and 100.
  • Inside 50, you will find a folder named 2024-1111-1139.
  • Inside 100, you will find a folder named 2024-1121-1355.
• Organize them into the project directory as follows:

CaDA
├── data
│   ├── lib_data
│   └── synthetic_data
├── 50
│   └── result
│       └── 2024-1111-1139
├── 100
│   └── result
│       └── 2024-1121-1355
└── utils

4. Prepare environment:

The project is developed with Python 3.8.15. Key packages include:

torch     2.0.1
torchrl   0.1.1
rl4co     0.2.0
tensordict                   0.1.2

The complete list of dependencies can be found in requirements.txt.

Training and Testing

For detailed instructions on training and testing the model, please refer to the README files inside the 50 and 100 directories.