How do I use the pretrained models to generate multiple images using a prompt?

[sarmientoj24]

The instructions in the Github is quite unclear.

[nanlliu]

You can modify demo.py to accomplish that. A prompt must be mapped to a label (see utility functions from our dataset class). Then u can stack the same labels for multiple times to use it as the final input label for generation. Hope that helps.

[sarmientoj24]

I still do not understand it. Say, I want to generate 1K images of 'a red square at the left of yellow ball'.

How do I do that?