Vision RAG
Attribution: Patterns inspired by Unsloth AI and Manuel Faysse Vision RAG tutorials.
VLM Architecture
A VLM combines:
- Visual encoder — ViT processes image patches
- Projection layer — maps visual tokens to LLM embedding space
- LLM decoder — generates text conditioned on visual + text tokens
Key VLMs (2024–2025):
| Model | Size | Strengths |
|---|---|---|
| Qwen2-VL | 2B, 7B, 72B | Multi-image, long context, document understanding |
| LLaVA-1.6 | 7B, |
[Description truncada. Veja o README completo no GitHub.]