Installation

The project is packed as a python package.

Use the following command to install it.

pip install -e .

If there are dependency issues, you may refer to the requirements.lock.txt. It is tested with Python 3.10.9, PyTorch 2.0.1, CUDA 11.7.