RAG App: A Customizable Retrieval-Augmented Generation Application

RAG App is a Python-based application designed to enhance natural language understanding by incorporating document (PDF/CSV) retrieval capabilities. It uses Retrieval-Augmented Generation (RAG) to answer queries based on private and previously unseen document collections, enabling accurate and contextual responses.

Installation

1. Clone the repository:

   git clone https://github.com/aliosmankaya/rag-app.git
   cd rag-app

2. Install dependencies using Poetry:

   poetry install

Project Tree

📦 rag-app
├─ .gitignore
├─ LICENSE
├─ README.md
├─ assets
│  └─ image.webp
├─ main.py
├─ poetry.lock
├─ pyproject.toml
├─ settings.toml
└─ src
   ├─ __init__.py
   ├─ api
   │  ├─ __init__.py
   │  ├─ db.py
   │  ├─ file.py
   │  ├─ model.py
   │  └─ search.py
   ├─ config.py
   ├─ core
   │  ├─ __init__.py
   │  ├─ db.py
   │  ├─ file.py
   │  └─ model.py
   └─ parser
      ├─ __init__.py
      ├─ db.py
      ├─ file.py
      ├─ model.py
      └─ search.py

Usage

1. Run the Application:

   python main.py

2. File Operations:

   • Upload File
     • POST /upload
       • Upload a file with multipart/form-data (PDF/CSV).
       • Response: { "Uploaded successfully." }
   • List Files
     • GET /list
       • Retrieve all uploaded files.
       • Response: [ *files ]
   • Update File
     • PUT /update
       • Update an existing file with multipart/form-data.
       • Response: { "Updated successfully." }
   • Delete File
     • DELETE /delete
       • Delete a file.
       • Response: { "Deleted successfully." }

3. Database Operations:

   • Create Vector DB Collection
     • POST /create
       • Create a vector database collection.
       • Response: { "Created successfully." }
   • Insert File to Collection
     • POST /insert
       • Insert embeddings of the file to the collection.
       • Response: { "Inserted successfully." }
   • List Collections
     • GET /list
       • Retrieve all uploaded collections.
       • Response: [ *collections ]
   • Update Collection
     • PUT /update
       • Update an existing collection.
       • Response: { "Updated successfully." }
   • Delete Collection
     • DELETE /delete
       • Delete a collection.
       • Response: { "Deleted successfully." }

4. Search:

   • Search relevant chunks (part of the file)
     • POST /search
       • Search with a query on the file.
       • Response: [ *chunks ]

5. Model:

   • Question answering powered by a LLM model using RAG
     • POST /model
       • Asking question about the file and answering by a LLM.
       • Response: { "..." }

Customization & Configuration

• Modify settings.toml to adjust retrieval or generation parameters.
• Update pyproject.toml for dependency management.

Contributions

Contributions are welcome! Please submit issues or feature requests via GitHub's issue tracker.

License

This project is licensed under the Apache 2.0 License.

---

Feel free to modify or add more specific instructions if needed!