Help training the network

Hi Adrian,

I run you network on some testing data and the results are amazing!!
I would like to train your network on another dataset with a different number of points.
Is it possible for you to supply the training code? This would great!

Thanks,
Alex

The issue requests training code for a custom dataset. The repository is a face alignment library built with PyTorch. To understand the existing training pipeline, examine the repository's scripts or examples directory. Look for existing training configurations or documentation on model architecture. The goal is to adapt the training process for a different number of input points.

索引モデルが issue の本文から書いたものです。