Foundation Model for Communication Systems Datagen

Overview

Code for generating the dataset for the paper "Towards a Foundation Model for Communication Systems".

arXiv Link: https://arxiv.org/abs/2505.14603

Citation:

@inproceedings{
  buffelli2025towards,
  title={Towards a Foundation Model for Communication Systems},
  author={Davide Buffelli and Sowmen Das and Yu-Wei Lin and Sattar Vakili and Chien-Yi Wang and Masoud Attarifar and Pritthijit Nath and Da-shan Shiu},
  booktitle={ICML 2025 Workshop on Machine Learning for Wireless Communication and Networks (ML4Wireless)},
  year={2025},
  url={https://openreview.net/forum?id=VZzF53BH0h}
}

Installation

Step 1: Create the Conda Environment

Create a new conda environment using the environment.yml file provided in the repository:

conda env create -f environment.yml

Step 2: Activate the Conda Environment

Activate the newly created conda environment:

conda activate venv

Step 3: Data Generation

To generate data run the following:

python generate_data/generate.py <path_to_config_directory> <output_folder>

Available options:

• overwrite: Overwrite existing saved data. If not specified, configs for which data exists in the output folder will be skipped.
• process: number of concurrent processes to use for multiprocessing. Too many will slow down the throughput
• files: exactly specify the name of a config file. Example: --files 0.yaml 1.yaml will run generation using the two specified files
• batch: Specify batch size. Default is 10

Config Generation

To generate config files run:

python generate_config_files.py <path_to_config_directory> -n 10

This will generate 10 random configurations in the directory provided. If n is greater than all possible combinations, we will generate all possible files.