Return Styles: Pseud0ch, Terminal, Valhalla, NES, Geocities, Blue Moon.

Pages: 1-

/pol/ mad about DeepSeek

Name: Anonymous 2025-01-24 10:59

https://boards.4chan.org/pol/thread/495245225#pc495245532
Every DEI FAANG company destroyed by 2000 lines of Chinese pytoddler code and $5.5M in training time. That's it, that's the whole thing.
UTTERLY EMBARRASSING!
FIRE EVERYONE in silicon valley, send ALL the jeets back

Name: Anonymous 2025-01-24 11:15

This isn't AGI. This is waste of GPU time on brute-force solutions.
Obvious there is some faster way that avoids the heavy math and storing all those huge matrices of tokens.
And why even use tokens? This forces continuations by token density by pure statistics, that like Markov chains deciding the next words being the most probable continuation by sheer volume.
Besides tokens make zero-sense for Chinese or any complex language where glyphs hold combinatory meanings(like programming object chains).

Name: Anonymous 2025-01-28 17:01

OpemAI engineers malding and quitting.
Meta assembles all code monkeys on hand to beat DeepSeek.
Its gonna be a long week.

Don't change these.
Name: Email:
Entire Thread Thread List