Multimodal AI ChatApp for diverse content types
Interact with PDFs, URLs, Youtube Videos, Images & Audios
- Built using Langchain, Streamlit
- Monitored via langsmith
- Models used and other configuration information can be found in config.yaml
Python Version: 3.10.12 Packages: Pls refer pip_freeze.txt and requirements.txt for package info
Project Information:
• A mutlisource multimodal AI chat assistant integrating Whisper AI for audio, LLaVA for images, Chroma DB for efficient PDF interaction, LangSmith for LLM Orchestration and observability, supporting 5 different data sources.
• Optimized the application to run quantized models, improving performance by 35% and reducing load times by implementing model caching and SQLite database for efficient management of chat history.