←back to thread

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

(github.com)
343 points sillysaurusx | 4 comments | 05 Mar 23 04:28 UTC | HN request time: 0.489s | source
1. sillysaurusx ◴[05 Mar 23 10:04 UTC] No.35028260[source]▶
Update: FB disabled the download link, so I mirrored everything to R2 and updated the script to use it. It should be working now (though the speed is "only" around 50MB/s).
replies(2): >>35030631 #>>35033063 #
2. toomuchtodo ◴[05 Mar 23 15:51 UTC] No.35030631[source]▶
Have you dropped the artifacts in the Internet Archive yet by chance?
replies(1): >>35032146 #
3. sillysaurusx ◴[05 Mar 23 18:01 UTC] No.35032146[source]▶
I'm surprised Internet Archive is appropriate for a 220GB model weight dump.

Please feel free; it seems like a good idea. I'm not sure I have enough weekend left to figure out yet another upload service today.

4. e12e ◴[05 Mar 23 19:33 UTC] No.35033063[source]▶
I've read the readme - but I'm not sure why this is any faster than just adding seeds to the torrent? More people downloading via torrent than http?