Adair Gerke

@sagerke

Joined July 2010

454 Following

50 Followers

1 Posts

sagerke retweeted

Magic @magicailabs

about 2 months ago

Frontier pretraining is said to be a big-lab-only game. We don’t have 100k chips yet, so there’s only one way: algorithmic efficiency. Our new recipe matches DeepSeek V4 Pro’s pretrain using 50x less compute – that’s roughly half the FLOPs used for GPT3, or ~$0.5M on GB200. https://t.co/uBs5o5Arsu

magicailabs's tweet photo. Frontier pretraining is said to be a big-lab-only game. We don’t have 100k chips yet, so there’s only one way: algorithmic efficiency. Our new recipe matches DeepSeek V4 Pro’s pretrain using 50x less compute – that’s roughly half the FLOPs used for GPT3, or ~$0.5M on GB200.

https://t.co/uBs5o5Arsu

179

468K

Adair Gerke

@sagerke

Last Seen Users on Sotwe

Trends for you

Most Popular Users