watsonx.ai Chat Application

A Streamlit-based chat application that interfaces with IBM's watsonx.ai foundation models, featuring document-based knowledge retrieval (RAG) capabilities.

Features

• 💬 Interactive chat interface with watsonx.ai models
• 📚 Document management with RAG (Retrieval-Augmented Generation)
• 🔧 Adjustable model parameters (temperature, tokens, etc.)
• 📊 Token usage tracking
• 📁 Support for multiple document formats (PDF, CSV, TXT, DOCX)
• 🔐 Secure credential management

Prerequisites

1. IBM Cloud Account with watsonx.ai access
2. API Key and Project ID from IBM Cloud
3. Python 3.8 or higher
4. pip package manager

Installation

1. Clone the repository:

git clone https://github.com/aseelert/watsonx-chatwithdocs
cd watsonx-chatwithdocs

2. Create a virtual environment (recommended):

python -m venv venv312
source venv/bin/activate  # On Windows: venv\Scripts\activate

3. Install required packages:

pip install -r requirements.txt

Configuration

1. Launch the application:

streamlit run Home.py

2. Navigate to the "Credentials" page
3. Enter your watsonx.ai credentials:
   • API Key
   • Project ID
4. Select your closest region
5. Verify the connection

Usage

Chat with watsonx.ai

Simple chat interface with RAG enabled.

use Document Knowledge

Chat Interface

• Navigate to the "Chat" page
• Adjust model parameters in the sidebar:
  • Temperature (0.0 - 1.0)
  • Max Tokens (100 - 4096)
  • Top P (0 or 1)
  • Top K (0 - 100)
  • Repetition Penalty (1 or 2)

Document Management

• Go to the "Document Management" page
• Upload documents (PDF, CSV, TXT, DOCX)
• Enable "Use Document Knowledge" in Chat to use RAG

File Structure

watsonx-chatwithdocs/
├── Home.py                 # Main application entry
├── pages/
│   ├── 01_Chat.py         # Chat interface
│   ├── 02_Documents.py    # Document handling
│   └── 03_Credentials.py  # Credential management
├── utils/
│   ├── watsonx_helpers.py  # watsonx.ai integration
│   ├── chroma_helpers.py   # Vector database management
│   └── credentials_helpers.py # Credential handling
└── requirements.txt        # Python dependencies

Dependencies

• streamlit
• langchain
• chromadb
• ibm-watsonx-ai
• python-dotenv
• sentence-transformers
• unstructured
• pypdf
• python-docx

Security Notes

• Credentials are stored locally in a SQLite database
• API keys are stored securely and not exposed in the UI
• Choose the closest region for optimal performance

Troubleshooting

1. If ChromaDB fails to initialize:

   • Ensure write permissions in the application directory
   • Check if required models are downloaded

2. If watsonx.ai connection fails:

   • Verify API key and Project ID
   • Check region selection
   • Ensure IBM Cloud account has necessary permissions

Contributing

Feel free to submit issues and enhancement requests!

License

MIT License