Boeun-Kim
Thank you for the great work! I'm trying to train your code with HumanML3D dataset and I have a simple question. How did you get the mirrored data from HumanML3D? I followed **raw_pose_processing.ipynb** in the repository of HumanML3D from top to bottom, but I just got npy files named with **6-digits numbers in 'joints' folder.** I confirmed that annotations.json in your repository includes data that their paths are like "**M**/CMU/80/80_32_poses". How did you get these mirrored motion files? Thank you!