zai-org/CogView
A 4-billion-parameter transformer model that generates images from text prompts, developed at Tsinghua University.
Not currently ranked — collecting fresh signals.
star history
CogView is a pretrained transformer for text-to-image generation across general domains. It supports Chinese text input (with translation from other languages supported) and introduced training techniques like PB-relax and Sandwich-LN for stabilizing large deep transformer training. The project provides pretrained models, inference code, and a demo interface for generating images from textual descriptions.
Frequently asked
- What is zai-org/CogView?
- A 4-billion-parameter transformer model that generates images from text prompts, developed at Tsinghua University.
- Is CogView open source?
- Yes — zai-org/CogView is open source, released under the Apache-2.0 license.
- What language is CogView written in?
- zai-org/CogView is primarily written in Python.
- How popular is CogView?
- zai-org/CogView has 1.8k stars on GitHub.
- Where can I find CogView?
- zai-org/CogView is on GitHub at https://github.com/zai-org/CogView.