We would like to create a pre-trained model in order to create a highly accurate model with less training data.Is it correct to use the checkpoint file (checkpoint_(data_dir_name)_(step).pt) generated during normal training?
· Sign up or log in to comment
