As long as I’m in a listy mood, this language model that backspaces out of dead ends is interesting because:
1. It’s an example of something categorically less interesting than the pro-LLM camp is going to claim, and more interesting than the anti-LLM camp wants to talk about; that gap seems important.
2. It points toward the text diffusion models I keep trying to subtly warn people are probably coming without saying so explicitly, oops, backspace backspace backspace.
ML nonsense
Went back and read the main DDPM paper to check for clues.
And found this spot on page 4 where they specifically say that what I put all that effort into trying is a bad idea.
So that’s neat.
ML nonsense
This hacky lightweight 4× super-resolver has learned that I want things to look sharp but not too sharp, so it sort of scatters edges around the image like spice in a sauce.
To my great though trivial annoyance, their processing pipeline cuts off the limb. It does look great, though, as expected. This is from the 4/20 eclipse, which @dscovr_epic also got an excellent frame of. Link to a 30 megabyte 1 km/pixel version: https://xoc.s3.amazonaws.com/also/gk2a-2023-04-20-0400.jpeg
Next time I write a CV it’s going to have a line like “I relish learning new datasets and working to make sense of them” but it’s going to be a lie because here we see longitude stored in radians:
I have a question about this, and the question is: WWWHHHHHHHAAAATTT?! https://iopscience.iop.org/article/10.3847/2041-8213/acba86
You know him on the internet. Eucalypt-adjacent; very occasional writer. Consulting and passively looking for work in geospatial, image processing, and related fields.