Improved ViT-VQGAN enables autoregressive Transformer pretraining on ImageNet tokens to reach IS 175.1 and FID 4.17 for generation plus 73.2% linear-probe accuracy, beating prior iGPT models.
Conditional image generation with pixelcnn decoders
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.CV 1years
2021 1verdicts
ACCEPT 1representative citing papers
citing papers explorer
-
Vector-quantized Image Modeling with Improved VQGAN
Improved ViT-VQGAN enables autoregressive Transformer pretraining on ImageNet tokens to reach IS 175.1 and FID 4.17 for generation plus 73.2% linear-probe accuracy, beating prior iGPT models.