This is the implementation of Text2Data on Motion dataset. The
code is adapted from the official code of MDM:
https://guytevet.github.io/mdm-page/

Environment:
python=3.7.16
pytorch=1.13.1
numpy=1.21.6
scipy=1.7.3
matplotlib=3.1.3
pandas=2.1.1
collections=3.10

Dataset can be collected from: https://guytevet.github.io/mdm-page/

To train unconditional model:
python train_uncond.py

To train EDM-finetune:
python train_ft.py

To train Text2Data:
python train_ft_lexico.py

Code regarding constraint optimization lies in "constraint"

Evaluation strictly follows the evaluation process in:
https://guytevet.github.io/mdm-page/
