Prompt-to-Prompt Cross-Attention Editing Quiz

Answers and explanations appear as you take the quiz.

1. What internal information does Prompt-to-Prompt reuse to preserve a scene?
   • The cross-attention maps linking words to image regions
   • The final output pixels
   • The model's optimizer state
   • The image's compression metadata
2. Does Prompt-to-Prompt require fine-tuning the model?
   • Yes, a full retrain per edit
   • No, it is a training-free technique
   • Yes, but only the text encoder
   • Only for adding new words
3. To swap a 'bicycle' for a 'motorcycle' while keeping the scene, the method:
   • Masks the bicycle region by hand
   • Trains a new LoRA for motorcycles
   • Maps attention between the corresponding tokens across the two prompts
   • Generates a completely independent image
4. What does reweighting a token's attention accomplish?
   • It changes the image resolution
   • It deletes the token entirely
   • It reorders the words in the prompt
   • It strengthens or weakens that word's visual influence
5. When you add a word to the prompt, Prompt-to-Prompt:
   • Preserves attention for unchanged tokens and lets only the new token form fresh attention
   • Re-randomizes the whole image
   • Ignores the added word
   • Switches to a different base model
6. Which separate model was trained to follow written image-editing instructions from an input image?
   • InstructPix2Pix, a conditional diffusion editor trained on generated instruction-edit pairs
   • Prompt-to-Prompt, which edits by controlling cross-attention maps during denoising
   • Null-text inversion, which optimizes unconditional embeddings for image inversion
   • LocalBlend, which creates a selected-token mask from attention maps
7. What does the Prompt-to-Prompt LocalBlend feature use to localize a change?
   • A hand-drawn mask required for every Prompt-to-Prompt edit
   • The EXIF metadata of the original image
   • Attention maps for selected prompt words to construct a spatial mask
   • A classifier that identifies image copyright
8. What extra step can enable Prompt-to-Prompt editing of a real photograph?
   • Train a new diffusion model from scratch
   • Convert the photograph into an audio spectrogram
   • Replace every cross-attention layer with a classifier
   • Invert the image into the diffusion model’s latent trajectory first