VTLWA Project

Vision-Language-Time-World-Action model deployment based on Gemma 4 E2B.

Structure

VTLWA/
├── pyproject.toml
├── README.md
├── src/
│   └── gemma4_e2b_inference.py   # Gemma 4 E2B inference script
└── cache/                         # HuggingFace model cache

Usage

# Download model (if not already cached)
HF_ENDPOINT=https://hf-mirror.com HF_HOME=./cache python src/gemma4_e2b_inference.py --download-only

# Run inference
HF_ENDPOINT=https://hf-mirror.com HF_HOME=./cache python src/gemma4_e2b_inference.py --prompt "Your prompt here"