Hi! I am newbie to text to human motion generation topic, and have few questions about evaluation.

• Train dataloader seems quite clear, but still couldn't understand valid dataloader including text information.
• Especially,
  • How w_vectorization file is organized? and what is this for?

  • momask-codes/models/t2m_eval_wrapper.py

    Line 5 in 94a6636

    def build_models(opt):

    this modules had been trained separately ?? Where this come from?