Implementation of NWT, audio-to-video generation, in Pytorch. The paper proposes a new discrete latent representation named Memcodes, which can be succinctly described as a type of multi-head hard-attention to learned memory (codebook) key/values. They claim the need for less codes and smaller codebook dimensions in order to achieve better reconstructions.
Features
- Implementation of NWT
- Audio-to-video generation
- For Pytorch
- Multi-head hard-attention to learned memory (codebook) key / values
- Smaller codebook dimension
- Achieve better reconstructions
License
MIT LicenseFollow NWT - Pytorch (wip)
You Might Also Like
With the world of work changed forever, it’s essential to manage your workplace and assets together to effectively create a high-performing environment. The Eptura experience combines the power of workplace management software with asset management, enabling you to effectively operate your building and facilitate hybrid work.
Rate This Project
Login To Rate This Project
User Reviews
Be the first to post a review of NWT - Pytorch (wip)!