Hacker News new | ask | show | jobs
by Palmik 59 days ago
I do not think this is a good prompt or useful benchmark, but nonetheless, it seems to work better for me: https://chatgpt.com/share/69e88a94-ded8-8395-b5dc-abceb2f44d...
1 comments
minimaxir 59 days ago
Huh, that is indeed better. If ChatGPT Images 2.0/gpt-2-image is more nondeterministic than usual, than that is in itself a useful data point.
link
Palmik 59 days ago
Did you enable thinking for your experiment? Are you sure you were on the 2.0 rather than 1.5 version?
link
minimaxir 58 days ago
That experiment image was directly through the API on high. (no Thinking parameter like the Web UI)
link