Pre-trained Language Models Do Not Help Auto-regressive Text-to-Image Generation.
Yuhui ZhangBrandon McKinzieZhe GanVaishaal ShankarAlexander ToshevPublished in: CoRR (2023)
Keyphrases
- language model
- image generation
- autoregressive
- pre trained
- information retrieval
- high resolution
- probabilistic model
- speech recognition
- text mining
- non stationary
- training data
- training examples
- mean shift
- sar images
- machine learning
- high quality
- image sequences
- point correspondences
- multi view
- prior knowledge
- feature space