logoalt Hacker News

Tejush • today at 5:19 PM • 1 reply • view on HN

Any guess on the pre-training tokens/flops they consumed?

Replies

lostmsu • today at 6:31 PM

Technical report has a graph vs Kimi K2 with 1e21 FLOPs (but they don't claim that's the entirety of pretraining)

➕ show 1 reply