---
title: NLP Assignment RAG App
emoji: 🧠
colorFrom: blue
colorTo: purple
sdk: gradio
sdk_version: "4.44.0"
app_file: main.py
pinned: false
python_version: "3.12"
app_build_command: uv sync
---

# NLP Assignment – Research Paper RAG Assistant

This app retrieves and summarizes academic papers related to your research query using a **Retrieval-Augmented Generation (RAG)** pipeline built with:

- **FAISS** vector search (`index.faiss`, `index.pkl`)
- **SentenceTransformer CrossEncoder** reranker
- **Google Gemini (via LangChain)** as the LLM
- **Gradio** UI frontend
- **uv** package manager (modern, fast, Python-native)

---

## 🚀 How to Run Locally

You can run this project locally using **uv** (recommended) or plain **Python**.

### 🧩 Prerequisites

- Python 3.12 or higher  
- [uv](https://docs.astral.sh/uv/) installed:

```sh
  pip install uv
```

1. Clone the repository

```sh
git clone https://huggingface.co/spaces/njayman/nlp-assignment
cd nlp-assignment
```

2. Add your FAISS index files

Make sure your FAISS index folder is structured like this:

```sh
faiss_index/
├── index.faiss
└── index.pkl
```

3. Set up environment

Set your Gemini API key (replace with your own key):

Linux/macOS:

```sh
export GOOGLE_API_KEY="your_google_api_key_here"
```

Windows (PowerShell):

```sh
$env:GOOGLE_API_KEY="your_google_api_key_here"
```

Or add a `.env` file

```sh
GOOGLE_API_KEY="your_google_api_key_here"
```

4. Create a virtual env and install dependencies using uv

```sh
uv venv
uv sync
```

5. Run the Gradio app

```sh
uv run main.py
```

Then open the link printed in your terminal — usually:

```sh
http://127.0.0.1:7860
```