README.md

---
title: AnythingQ&A
emoji: ❓
colorFrom: blue
colorTo: red
sdk: gradio
sdk_version: 3.48.0
app_file: app.py
pinned: false
---

## Anything Q&A

This tool serves as a customizable version of the amazing ClimateQA by Ekimetrics. It allows you to rapidly create a new question-answering tool using any set of documents as a data source.

## Quick Start

0. **Fork the Repository**  
   Fork the original HuggingFace space: [Anything Q&A on HuggingFace](https://huggingface.co/spaces/LouisSanna/anything-question-answering).

1. **Set environment variables**

For OpenAI API:

```env
OPENAI_API_KEY=...
```

2. **Install Dependencies**

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

3. **Add your data.**
   Place the PDFs to be used as sources in the data folder. The subfolder and file names will be used as default identifiers by the tool.

   ```txt
   data/
       type_1/
           source_1.pdf
           source_2.pdf
       type_2/
           source_3.pdf
   ```

4. **Build the index of semantic vectors**

   ```bash
   python -m anyqa.build_index
   ```

5. **Launch the app**

   ```shell
   python app.py
   ```

And you're done!

## Deployment

Deploying via Hugging Face Spaces is the most straightforward approach. Simply push your code to a HuggingFace Gradio space, and it will function as is.

You will need to set the environments variables in the settings.

## AI providers

The following providers are currently supported:

- OpenAI
- Azure OpenAI

The provider is picked based on the variable variables, see `example.env`.