Glow: Better reversible generative models
OpenAI Blog
Glow introduced an invertible generative model using invertible 1x1 convolutions that simplifies prior reversible architectures. The model generates high-resolution images while supporting efficient sampling and enables attribute manipulation through learned feature representations. Researchers released open-source code and an online visualization tool enabling others to build on the approach.
Why it matters
We introduce Glow, a reversible generative model which uses invertible 1x1 convolutions. It extends previous work on reversible generative models and simplifies the architecture. Our model can generate realistic high resolution images, supports efficient sampling, and discovers features that can be used to manipulate attributes of data. We’re releasing code for the model and an online visualization tool so people can explore and build on these results.