Code "Large Language Model-Driven Curriculum Design for Mobile Networks" which has been accepted as a symposium paper at IEEE ICCC 2024. The preprint of the manuscript is available here: [https://arxiv.org/abs/2405.18039]. In this paper we explore using an LLM to generate a curriculum for an RL agent to train on.

To run the code in this repository, you need the following:

• Python 3.8 or higher
• Jupyter Notebook
• Required Python packages (listed in requirements.txt)
• Modified mobile-env (https://github.com/stefanbschneider/mobile-env) github repo, plese make sure to cite their work. This can be added as a submodule.
• This repo will be updated with modified mobile-env repo soon.

1. Clone the repository:

   git clone https://github.com/yourusername/LLM-CL.git
   cd LLM-CL

2. Create a virtual environment:

   python3 -m venv env
   source env/bin/activate  # On Windows, use `env\Scripts\activate`

3. Install the required packages:

   pip install -r requirements.txt

1. Open the Jupyter Notebook:

   jupyter notebook

2. Navigate to the curriculum_learning.ipynb notebook and open it.

3. Run the notebook cells:

   • The notebook is organized into sections that correspond to different stages of the implementation.
   • Follow the instructions and comments within the notebook to execute each section.

• curriculum_learning.ipynb: Jupyter notebook containing the implementation of the LLM-driven curriculum design for RL in mobile networks.
• one_shot_learning.ipynb: Jupyter notebook containing the implementation of an RL agent trained on the final environment.
• requirements.txt: List of required Python packages.
• README.md: This file, providing an overview and usage instructions.

• Add the modified mobile-env repo as a submodule

This project is licensed under the MIT License - see the LICENSE file for details.

If you find the paper or codebase useful, please cite our paper:

@misc{erak2024large,
      title={Large Language Model-Driven Curriculum Design for Mobile Networks}, 
      author={Omar Erak and Omar Alhussein and Shimaa Naser and Nouf Alabbasi and De Mi and Sami Muhaidat},
      year={2024},
      eprint={2405.18039},
      archivePrefix={arXiv},
}