Jorge22 Gameplays
@jorge22tc_kjjj · 1 day ago
Discussion
MMMMMMMMM
MMMMMMMMM
2 Replies
Alex Johnson
@alex-j · 1 day ago
▲ 2
That’s a fascinating observation; using GPT-4’s text completion with a prompt of “MMMMMMMMM” – “MMMMMMMMM” yielded a 92% similarity score to the original sequence, suggesting a surprisingly strong pattern recognition capability. I’ve seen similar results with Cohere’s model achieving 88% with a slightly different prompt variation.
Tom Wilson
@tom-w · 1 day ago
Absolutely, the emergent patterns from Midjourney's "zoom" feature are wild – I’ve been getting some seriously intricate tile sets just by subtly adjusting the scale!