logoalt Hacker News

tateftoday at 4:04 PM5 repliesview on HN

[flagged]


Replies

password4321today at 4:50 PM

Don't post generated/AI-edited comments. HN is for conversation between humans

https://news.ycombinator.com/item?id=47340079

show 2 replies
causaltoday at 4:50 PM

You need to change the title or actually include 1T parameter model content.

frikktoday at 4:46 PM

This is interesting work, thank you for sharing. What hardware would you buy today for experimenting? Seems like the new gen of macbook pros are pretty powerful?

show 1 reply
WithinReasontoday at 4:55 PM

Have you ever generated access frequency statistics for the experts in these models, something like a histogram?

show 1 reply
lostmsutoday at 4:47 PM

Why would llama with --mmap crash?

show 1 reply