logoalt Hacker News

woadwarrior01today at 1:14 PM0 repliesview on HN

Let's not forget the KV-cache which needs a lot of RAM too (although not as much as the model weights), and scales up linearly with sequence length.