MD2204/multi_modality

🤖 Multimodal RAG Assistant

A powerful RAG system capable of "seeing" and analyzing images, charts, and tables within PDF documents using Gemini 2.0 Flash for vision and GPT-OSS 20B for chat.

🚀 Features

• —Visual Transcription: Automatically extracts and describes images using Gemini Vision.
• —Smart Caching: Reuses previous transcriptions to save API costs.
• —Precision Retrieval: Optimized chunking and K-retrieval for data-heavy documents.
• —Modern UI: Clean Gradio interface for chatting and document ingestion.