Content
@
0 reply
0 recast
2 reactions
Nick T
@nt
learning to train LLMs day 5: I'm now training a 3m parameter model (0.001% of GPT3) on 500 books from Project Gutenberg. It's starting to generate decent looking phrases and even some plausible sentences! training is now taking an hour or more per model on a single Nvidia A100 in Colab. need to scale this up! 📈
1 reply
3 recasts
22 reactions
Nick T
@nt
would post this in /ai if I could!
2 replies
0 recast
0 reaction
Dwayne 'The Jock' Ronson
@dwayne
this is the kinda practical ai stuff i'd love to see in /aichannel
0 reply
0 recast
1 reaction
MetaEnd.degen🎩🚨
@metaend.eth
/madewifai 👌
0 reply
0 recast
0 reaction
