CF_VLA

CF_VLA is a two-stage VLA (Vision-Language-Action) training and inference project.

Core Information

• Core model implementation: src/openpi/models_pytorch/cf_vla.py
• Compatible configuration name: pi0_2stg_pytorch
• Training entry point: scripts/train_pytorch.py
• One-click launch script: scripts/run_train_cf_vla.sh

Environment Setup

Use the standalone environment file environment.yml to create the cf_vla conda environment for GPU reproduction.

# 1. Create the environment
conda env create -f environment.yml

# 2. Activate the environment
conda activate cf_vla

# 3. Install the project package
pip install -e .

# 4. Install the openpi client
pip install -e packages/openpi_client

Optional Benchmark Dependencies

pip install -e ".[bench-libero]"
pip install -e ".[bench-calvin]"

Training Example

bash scripts/run_train_cf_vla.sh \
  configs/pi05_2stg_pytorch_delta_actions/Cf_vla_libero.yaml

Inference and Evaluation Example

# Server side
python scripts/serve_policy.py --env LIBERO

# Client side
python examples/libero/main.py
python examples/calvin/main.py

License

• Code license: MIT. See LICENSE.
• Third-party dependency notes: THIRD_PARTY_NOTICES.md
• Model weights and datasets may have separate licenses. Please verify them separately before release.