VerMind Family

Next-generation AI models designed for language understanding, multimodal reasoning, and efficient deployment.

Open Source • MIT License

🎯 Interactive Web Demo

Try VerMind models directly in your browser with our Gradio-powered interface

LLM Mode - Pure Text Conversation

LLM Demo
python3 scripts/web_demo.py --model_path nev8r/vermind --mode llm

VLM Mode - Vision + Language

VLM Demo
python3 scripts/web_demo.py --model_path /path/to/vermind-v --mode vlm
🧠
Language

VerMind

Foundation Language Model

High-performance decoder-only transformer with GQA, SwiGLU, and RoPE. Optimized for efficient training and inference.

  • Grouped Query Attention
  • 32K context length
  • DPO / PPO / GRPO
  • LoRA fine-tuning
Explore VerMind
👁️
Vision + Language

VerMind-V

Vision-Language Model

Multimodal extension with vision encoder integration. Understand and reason about images alongside text.

  • Vision encoder
  • Image captioning
  • Visual QA
  • Unified architecture
Explore VerMind-V