From dc61dc0feb599232454f1b8730a1129edcd6c576 Mon Sep 17 00:00:00 2001 From: Shawn Presser Date: Sun, 5 Mar 2023 02:44:22 -0600 Subject: [PATCH] Update README.md --- README.md | 2 ++ 1 file changed, 2 insertions(+) diff --git a/README.md b/README.md index b96a34b..328da3b 100644 --- a/README.md +++ b/README.md @@ -1,5 +1,7 @@ # llama-dl +**UPDATE (2:43 AM CST)**: Facebook has closed off this download vector. I'm currently mirroring the model to Cloudflare R2, and I'll update the script to use it right now; I'll keep you updated as I go. Check back in like... an hour? + This repository contains a high-speed download of LLaMA, Facebook's 65B parameter model that was recently made available via torrent. (Discussion: [Facebook LLAMA is being openly distributed via torrents](https://news.ycombinator.com/item?id=35007978)) It downloads all model weights (7B, 13B, 30B, 65B) at around 200 MB/s: