Anytime importing captions, the script will execute training on the first image and only the first image with the set number of steps then exit. The presence of a matching .txt file to the image seems to prevent it from iterating onto the next image or even tallying multiple images as part of the dataset.
Anytime importing captions, the script will execute training on the first image and only the first image with the set number of steps then exit. The presence of a matching .txt file to the image seems to prevent it from iterating onto the next image or even tallying multiple images as part of the dataset.
It seems to count the .txt as images?