Embed Notice
HTML Code
Corresponding Notice
- Embed this notice
iced depresso (icedquinn@blob.cat)'s status on Saturday, 03-Aug-2024 02:39:06 JSTiced depresso @shibao @sun i did have a sweet paper on using gradient descent to train those old gaussian mixture models way way better than they used to be, though, which would be kinda fun to slap on a text to speech system because those things ran on potatos.