Deeplabcut pretained model for horses and extracting keypoints?

Hi,
Very new to this community. I have a basic knowledge of computer vision .I had been learning primarily by experimenting with 2D human pose estimation models such as openpose, posenet. I read that there was a pre trained model for keypoints for single horse where it moves from left to right ? I was wondering if there is a guide on

  1. How it can be setup ?
  2. Run the pre trained model on other videos (horses moving from left to right )
  3. How can one access ,its specific keypoints? (In 2d human pose estimation, they are marked as left_eye,right eye etc)
    Hope to learn and hear from you
    Vis

The following link should be useful for you to get started: