submitted 3 days ago by DamnianWayne@lemmy.world to c/world@quokk.au

18 comments fedilink hide all child comments

you are viewing a single comment's thread
view the rest of the comments

[-] Randomgal@lemmy.ca 2 points 3 days ago

Lol what? They already did.

[-] nymnympseudonym@piefed.social 2 points 3 days ago

Not ones to compete with NVidia Not ones that will train GPT5/6/7

[-] cassandrafatigue@lemmy.dbzer0.com 3 points 3 days ago

The.... That's not

Enough data to train a hypothetical gpt6 literally does not exist.

[-] nymnympseudonym@piefed.social 0 points 3 days ago

Most frontier models since GPT4 actually use synthetic pre-training data , before slurping up the Internet and after along with RLHF

[-] cassandrafatigue@lemmy.dbzer0.com 2 points 3 days ago

~~too many fingers~~ sausage fingers

[-] nymnympseudonym@piefed.social 1 points 3 days ago

Also training data is totally independent of chip feature size, interconnects, memory bandwidth, and all the things actually relevant to chips

this post was submitted on 09 Dec 2025

75 points (100.0% liked)

World News

1020 readers

475 users here now

Rules:
Be a decent person, don't post hate.

Other Great Communities:

Rules

Be excellent to each other

founded 1 year ago

MODERATORS