Congratulations on your work!Very exciting and it seems to emphasize the importance of optimized noise initialization. I would like to know the process of getting the target token for any prompt I would like to give. The paper states "manually or automatically with GPT", could you expand more on the process please, are they just the indices on the text prompt I would like to emphasize?
Congratulations on your work!Very exciting and it seems to emphasize the importance of optimized noise initialization. I would like to know the process of getting the target token for any prompt I would like to give. The paper states "manually or automatically with GPT", could you expand more on the process please, are they just the indices on the text prompt I would like to emphasize?