ahatamiz opened this issue a year ago · 1 comments
Based on conversation in #56 , it is a good feature to allow running HAT also from a separate flag to avoid adding new layers when the goal is to only use a different resolution for inference.
A new MR could address this.
Addressed by #60