Scott H. Hawley

172 Followers
72 Following
89 Posts

Interests:
Machine Learning + Musical Audio Production. Numerical modeling, Teaching audio engineers, Tech & Society, Science & Religion. Former astrophysicist.

Affiliations:
- Professor of Physics, Belmont University
- Senior Fellow, Belmont Data Collaborative
- Formerly: Technical Fellow, Harmonai (Stability AI)

Websitehttps://hedges.belmont.edu/
GitHubhttps://github.com/drscotthawley/
LinkedInhttps://www.linkedin.com/in/drscotthawley/
"AI Art"https://www.instagram.com/xaltint/
Ooopsie!
Oi mate that Python code segment is def __init__?
...but probably won't get around to.
We're the same kind of weird if:
You text yourself links to papers you want to read.
Now on BlueSky too. https://bsky.app/profile/drscotthawley.bsky.social (Not sure which of these alt-Twitters I'll frequent.)
Scott H. Hawley (@drscotthawley.bsky.social)

Prof. of Physics at Belmont University, Teacher of audio engineers, ML enthusiast

Bluesky Social
Correct me, but:
Vector-Quantized 'Variational' AutoEncoders (VQ-VAEs) are not variational. We should just call them VQ-AEs. There is no mean or variance, there is no sampling. They are completely deterministic.
I got tired of seeing posts about "10 Reasons Why X is a Christmas Movie", so I made into a game for ChatGPT. https://chat.openai.com/share/f7b52f93-692a-4113-a5cd-7b6df25cde5d
Deep Thought paused for a moment's reflection.
"Tricky," he said finally.
You are to respond like a know-it-all guy. Please begin every response from now on with "Well, actually". Go.
Book proposal on classification is coming along well. Today I added 1800 words plus drafted this figure showing forms we'll cover. Please come at me now; it's imperfect & I welcome suggestions.
(Note that you're missing my paragraph of disclaimers. ;-) ) #classification