SAGE

SAGE: Data-Semantics-Aware Recommendation of Diverse Pivot Tables

This repository contains the implementation of SAGE, Data-Semantics-Aware Recommendation of Diverse Pivot Tables.

---

📦 Requirements

First, set up your Python environment and install the dependencies.

Use requirements.txt

pip install -r requirements.txt

---

🤗 Hugging Face Model Access

SAGE uses models like LLaMA, TAPEX, and T5 hosted on Hugging Face. Access to these models may be restricted.

Steps:

1. Request access to the model by following the link.
2. Once approved, create a Hugging Face token from your settings page.
3. Save your token in a file named hf_token.txt in the root directory.

---

🚀 How to Run

After setting up dependencies and your token, you can run the main script

bash run_sage.sh

---

📂 Project Structure

├── hf_token.txt             # Hugging Face access token
├── requirements.txt         # Python dependencies
├── run_sage.sh              # The script to run the SAGE algorithm
├── src/                     # Source code modules
├── dataset/                 # Input datasets 
├── precompute/              # Precomputed models and results
└── README.md