Code World Model:
Coding Agent as World Brain
Yiwen Chen1,2 · Guosheng Lin2 · Chi Zhang1
1 AGI Lab, Westlake University 2 Nanyang Technological University
https://github.com/user-attachments/assets/4c13c5fb-ab45-4a0e-9e59-b85499ef7390
Quick start
- Follow the environment and installation guide.
- From the repository root, download the public CWM LoRA and deterministically expand the 40 compact Hugging Face examples:
./scripts/install_release_assets.sh "$PWD/release"
- Enter the prepared release directory and run a released configuration without changing its inputs, prompts, or seed:
cd release
CONFIG=examples/example_01/config.json
python -m cwm_h3_inference validate --config "$CONFIG"
CUDA_VISIBLE_DEVICES=0 python -m cwm_h3_inference prepare --config "$CONFIG"
CUDA_VISIBLE_DEVICES=0 python -m cwm_h3_inference generate --config "$CONFIG"
Use a new cache directory and output path for every changed experiment. The Hugging Face examples contain compact NPZ conditions, not caches or output videos; the installation script performs the required local format expansion.