logoalt Hacker News

nippootoday at 6:43 AM4 repliesview on HN

They've taken it down now and replaced with an arguably even less helpful diagram, but the original is archived: https://archive.is/twft6


Replies

yoz-ytoday at 7:33 AM

Wow it’s even worse than I thought. I thought that convictungly morhing would be the only problem. The nonsense and inconsistent arrowheads, the missing annotations, the missing bubbles. The “tirm” axis…

That this was ever published shows a supreme lack of care.

show 4 replies
fooyctoday at 1:45 PM

Apparently the new diagram is now a rip off of another one from Atlassian: https://bsky.app/profile/vurobinut.bsky.social/post/3mf52hmw...

franktankbanktoday at 1:39 PM

TIMMMAYYY

rzmmmtoday at 8:24 AM

It looks like typical "memorization" in image generation models. The author likely just prompted the image.

The model makers attempt to add guardrails to prevent this but it's not perfect. It seems a lot of large AI models basically just copy the training data and add slight modifications

show 2 replies