James Betker

A few interesting ways to understand how diffusion models work: nonint.com/2022/10/31/gro…

I just posted a write-up on my deep-learning hardware: nonint.com/2022/05/30/my-…

I released TorToiSe v2.3 tonight. The main new feature is a new CLVP model. 6x more parameters, 20% better performance on the validation set. TorToiSe itself performs ~5% better by github.com/neonbjb/tts-sc…

Friends don't let friends train small diffusion models: nonint.com/2022/05/04/fri…

Tortoise v2 is out! github.com/neonbjb/tortoi…

github.com/neonbjb/tortoi… a DALL-E like text-to-speech model trained on 16M speech samples.

I'd like to share ocotillo - an extremely simple speech to text library that uses wav2vec. Github: github.com/neonbjb/ocotil… Blog post: nonint.com/2022/01/22/bat…