Skip to content

RAG-nificent v0.0.4

Latest
Compare
Choose a tag to compare
@MaxMLang MaxMLang released this 06 Mar 22:59
· 12 commits to master since this release

🚀 New Features

Model Support

  • Added support for Llama-3.3-70b-versatile - Latest version of Meta AI's Llama model with enhanced capabilities
  • Added support for GPT-4o - OpenAI's latest multimodal model with improved reasoning capabilities
  • Added support for GPT-4o-mini - More efficient version of GPT-4o, now set as the default model
  • Added support for Gemma2-9b-it - Google's efficient instruction-tuned language model

User Interface

  • Follow-up Question Buttons - The system now generates and displays relevant follow-up questions as clickable buttons after each response
  • Improved Citations - Enhanced citation format with direct links to the specific page in source documents
  • Dockerized Application - Added Docker support for easier deployment and consistent environment

🔧 Technical Improvements

Application Architecture

  • Fallback Model System - Implemented cascading fallback to alternative models if the selected model is unavailable
  • Improved Error Handling - Added comprehensive error handling and logging throughout the application
  • Structured Output Format - Responses now include structured follow-up questions using Pydantic models and JSON output parsing

Code Quality

  • Enhanced Logging - Added detailed logging with both file and console output for better debugging
  • Modular Design - Refactored code for better separation of concerns and maintainability
  • Improved Exception Handling - Better error recovery mechanisms to ensure application stability

📦 Dependency Updates

  • Updated Pinecone client to the latest version with new initialization pattern
  • Updated LangChain components to latest versions
  • Added support for Groq API integration
  • Updated Chainlit UI components

🐛 Bug Fixes

  • Fixed vector store initialization issues with better fallback mechanisms
  • Resolved model loading errors with cascading fallback options
  • Improved handling of API rate limits and timeouts

📝 Documentation

  • Added Docker setup instructions to README
  • Updated model descriptions and capabilities
  • Added development workflow documentation
  • Updated license to Creative Commons Attribution-NonCommercial 4.0