Gids-gekoppelde quiz · Middelmatig Niveau
Quiz over vijandige diffusiedestillatie en turbomodellen
Test hoe tegenstanders en destillatieverliezen van leraren samenkomen in ADD, hoe LADD verschilt, en welke turbomodellen opgeven in diversiteit en controle.
Vraag 1 van 8
Welke twee verliezen combineert ADD?
Bekijk elke vraag8 vragen
Antwoorden en uitleg verschijnen terwijl je de quiz doet.
- Welke twee verliezen combineert ADD?
- An FID loss and a CLIP loss
- An adversarial loss and a distillation loss from a teacher
- A perceptual loss and a pixel loss only
- A contrastive loss and cross-entropy
- Welk model is naast het ADD-papier uitgebracht?
- Flux.1 dev
- DALL-E 3
- SDXL Turbo
- Stabiele diffusie 1.5
- Welke ruggengraat gebruikt de discriminator van ADD?
- A frozen DINOv2 feature network with trainable heads
- A fully trainable Inception network
- The teacher's text encoder
- No backbone; it compares raw pixels
- Waarin verschilt LADD van ADD?
- It uses more denoising steps
- It removes the teacher
- It only works for video
- It runs the discriminator in latent space using the teacher's features
- Waarom hebben negatieve aanwijzingen vaak weinig effect op turbomodellen?
- The tokenizer filters them out
- The models have no text encoder
- Classifier-free guidance is typically built in during distillation
- Negative prompts only work at 1024x1024
- What happens if you run a normal, undistilled diffusion model at one step?
- It matches SDXL Turbo
- It usually produces a blurry, unfinished image
- It generates video
- It refuses to run
- Which quality trade-off is common in distilled few-step models?
- Higher diversity than the teacher
- Better text rendering in every case
- Support for unlimited resolution
- Less variety across different seeds
- In ADD's distillation loss, what is the student's target?
- The teacher's denoised prediction of the re-noised student output
- A random real image
- The discriminator's logits
- The CLIP score of the prompt