←back to thread

Show HN: Llama-dl – high-speed download of LLaMA, Facebook's 65B GPT model

(github.com)
343 points sillysaurusx | 1 comments | 05 Mar 23 04:28 UTC | HN request time: 0s | source
Show context
arjvik ◴[05 Mar 23 11:12 UTC] No.35028574[source]▶
How big is this model? (i.e. disk space to store it)
replies(1): >>35028578 #
kuroguro ◴[05 Mar 23 11:14 UTC] No.35028578[source]▶
65B is ~120GB. All of them combined with the smaller versions is ~220GB.
replies(1): >>35030238 #
nextaccountic ◴[05 Mar 23 15:12 UTC] No.35030238[source]▶
After converting to int8, does it become smaller? Also, can this be further compressed? Like, is there some redundancy a special-purpose compressor could exploit?
replies(1): >>35030409 #
1. rfoo ◴[05 Mar 23 15:27 UTC] No.35030409[source]▶
Converting to int8 halves the size.