README.md

---
title: RAG
emoji: ⚡
colorFrom: red
colorTo: green
sdk: gradio
sdk_version: 6.1.0
app_file: app.py
pinned: false
license: apache-2.0
---

Check out the configuration reference at https://huggingface.co/docs/hub/spaces-config-reference
# File: `README.md`

```markdown
# RAG Document Assistant

A complete Retrieval-Augmented Generation (RAG) Python project using:
- HuggingFace Transformers
- sentence-transformers (for embeddings)
- ChromaDB (vector store)
- Gradio (UI)

## Project Structure
```

project/
├── app.py
├── rag_pipeline.py
├── generator.py
├── utils.py
├── requirements.txt
├── README.md
├── data/
└── db/

````

## Installation
1. Create a virtual environment (recommended):

```bash
python -m venv venv
source venv/bin/activate  # on Windows use venv\Scripts\activate
````

2. Install requirements:

```bash
pip install -r requirements.txt
```

3. Run the app:

```bash
python app.py
```

Open the Gradio URL shown in the console (default `http://127.0.0.1:7860`).

## How RAG Works (short)

1. Documents are uploaded and their text is extracted.
2. Text is chunked into overlapping passages.
3. Each chunk is embedded using a pretrained sentence-transformer.
4. Chunks and embeddings are stored in a vector database (ChromaDB).
5. At query time, the user question is embedded and used to retrieve most relevant chunks.
6. Retrieved chunks are passed to a generator LLM which composes a grounded answer.

## Notes & Troubleshooting

* Textract may need system-level dependencies for PDF/DOCX parsing on some platforms.
* Large models may require GPUs. For local CPU usage, prefer small models like `flan-t5-small`.
* If you see memory errors, reduce model size or run on a machine with more RAM.

## Roadmap / Improvements

* Add user authentication and per-user collections
* Support incremental indexing and deletion
* Add streaming generation for long answers
* Add API endpoints via FastAPI