Our code is based on Glow and MoGlow.
The conda environment defined in 'environment.yml' contains the required dependencies.
Our training data is available here.
Our pretrained model is available here.
Prepare the environment by: conda env create -f environment.yml
Download and extract the data and pretained model to the project.
For training:
python train.py hparams/locomotion.json locomotion.
For style transfer, , and then run:
python train.py hparams/locomotion_test.json locomotion
Please see the included LICENSE file for licenses and citations.