Update README.md

README.md

@@ -10,10 +10,10 @@ tags:
 
 ## Introduction
 
-Omnivision is a compact, sub-billion (968M) multimodal model for processing both visual and text inputs, optimized for edge devices. Built on LLaVA's architecture, it features:
+Omnivision is a compact, sub-billion (968M) multimodal model for processing both visual and text inputs, optimized for edge devices. Improved on LLaVA's architecture, it features:
 
 - **9x Token Reduction**: Reduces image tokens from 729 to 81, cutting latency and computational cost.
-- **Minimal-Edit DPO**: Enhances response quality with minimal edits, preserving core model behavior.
+- **Trustworthy result**: Reduces hallucinations using **DPO** training from trustworthy data.
   
 **Quick Links:**
 1. Interactive Demo in our [Hugging Face Space](https://huggingface.co/spaces/NexaAIDev/omnivlm-dpo-demo).