WiLL: Your FCC Expert

Welcome to the WiLL, an FCC expert system leveraging Large Language Models (LLMs) to answer questions on wireless communications and FCC regulations. This research was presented at HotMobile 2025 under the title:
"Can We Make FCC Experts Out of LLMs?"

Authors:
Atul Bansal, Veronica Muriga, Jason Li, Lucy Duan, and Swarun Kumar
Carnegie Mellon University

---

🚀 Getting Started

Step 1: Clone the Repository

git clone https://github.com/Jasonic121/WiLL.git
cd WiLL

Step 2: Set Up the Virtual Environment and Install Dependencies

python -m venv venv
source venv/bin/activate  # On Windows: venv\Scripts\activate
pip install -r requirements.txt

Step 3: Configure Environment Variables

Create a .env file in the root directory and set the necessary environment variables.

Step 4: Run the Terminal Interface

python WiLL_terminal.py

Step 5: Ask Questions and Explore FCC Knowledge!

Start querying the system to get insights into FCC regulations and wireless communication topics.

---

📂 Project Structure

WiLL-System/
├── WiLL_terminal.py      # Main terminal interface for the WiLL system
├── AblationStudyExp/     # Experimental setups for ablation studies
│   ├── Baseline/         # Baseline experiments with various models (GPT, LLaMA)
│   ├── Chunking/         # Text chunking strategy experiments
│   └── SimilarityFunction/ # Studies on similarity measurement approaches
├── csv_files/            # Data files for experiments
├── Finetune/             # Model fine-tuning scripts and configurations
├── environment/          # Environment setup and configuration files
└── outputEvaluation/     # Experimental results and evaluation metrics

---

📊 Ablation Studies

The AblationStudyExp/ directory contains various experiments designed to evaluate the performance of the WiLL system, including:

• Baseline Experiments: Comparing different LLMs (GPT, LLaMA) to assess baseline performance.
• Chunking Strategies: Evaluating the impact of different text chunking approaches on retrieval performance.
• Similarity Functions: Experimenting with various similarity measures to improve knowledge retrieval accuracy.

---

🔍 Environment Configuration

The project uses environment variables to manage API keys, paths, and other settings. Ensure that all required variables are defined in the .env file.

Example:

OPENAI_API_KEY=your_api_key_here
FINETUNE_KEY=your_finetune_key_here

---

🛠️ Development and Customization

Feel free to extend and customize the project as needed. Key areas for potential modification include:

• Model Fine-tuning: Enhance model accuracy by fine-tuning with domain-specific data in the Finetune/ directory.
• Ablation Studies: Experiment with different configurations to further optimize system performance.
• Evaluation Metrics: Customize evaluation scripts in outputEvaluation/ to suit your research needs.

---

📝 Citation

If you use this system in your research, please cite:

@inproceedings{WiLL2025,
  author    = {Atul Bansal and Veronica Muriga and Jason Li and Lucy Duan and Swarun Kumar},
  title     = {Can We Make {FCC} Experts Out of {LLMs}?},
  booktitle = {Proceedings of the 26th Annual International Workshop on Mobile Computing Systems and Applications (HotMobile 2025)},
  year      = {2025},
  publisher = {ACM},
  address   = {New York, NY, USA},
}

---

📬 Contact

For questions or support, reach out to the authors via their Carnegie Mellon University emails.

---