VQ-VAE ati Idanwo Latents ọtọtọ

Awọn idahun ati awọn alaye han bi o ṣe n mu ibeere naa.

1. Kini o jẹ ki aaye wiwaba VQ-VAE yatọ si ti VAE boṣewa kan?
   • O ti wa ni ọtọ, lilo a kẹkọọ codebook ti awọn ifibọ
   • O ti wa ni lemọlemọfún ati Gaussian
   • Ko ni aaye wiwaba rara
   • O tọju awọn piksẹli aise
2. Bawo ni VQ-VAE ṣe kọja awọn gradients nipasẹ igbesẹ titobi ti kii ṣe iyatọ?
   • O skips backpropagation šee igbọkanle
   • Pẹlu iṣiro-taara ti o daakọ awọn gradients kọja quantizer
   • Nipa lilo ẹkọ imuduro nikan
   • Nipa yiyi gradients si odidi
3. Kí ni 'codebook Collapse'?
   • Oluyipada ko foju pa awọn latents
   • Awọn abajade kooduopo di ailopin
   • Nikan ida kan ti awọn titẹ sii codebook lailai ni lilo
   • Ipinnu aworan lọ silẹ si odo
4. Kini idi ti awọn latents ọtọtọ wulo fun awoṣe ipilẹṣẹ pẹlu Awọn Ayirapada?
   • Wọn ṣe awọn aworan kere si lori disk nikan
   • Wọn ṣe imukuro iwulo fun decoder
   • Wọn ti yọ gbogbo randomness lati iran
   • Wọn yi media pada si awọn ilana ami ami Ayipada kan le ṣe awoṣe bi ede
5. Kini VQGAN ṣafikun lori oke ti ohunelo VQ-VAE ipilẹ?
   • Adversarial ati perceptual adanu fun ndinku reconstructions
   • Ayẹwo itankale
   • Yiyọ ti awọn codebook
   • Ayipada loorekoore
6. Oro isonu wo ni o tọju kooduopo 'ifaramo' si awọn titẹ sii koodu koodu ti o yan?
   • Iyipada ninu owo-owo KL
   • The ifaramo pipadanu
   • Awọn agbelebu-entropy pipadanu
   • Ipadanu meteta
7. What kind of prior does the original VQ-VAE paper learn over its discrete codes?
   • A fixed prior that is never learned
   • A prior over camera locations only
   • A learned prior over the discrete code sequence
   • No prior or decoder distribution
8. Which text-to-image paper used a discrete VAE to tokenize images before autoregressive generation?
   • The original Whisper speech-recognition paper
   • A paper on Kalman-filter vehicle tracking
   • The original BERT masked-language-model paper
   • The original DALL·E paper