←back to thread

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

(github.com)
343 points sillysaurusx | 2 comments | 05 Mar 23 04:28 UTC | HN request time: 0s | source
Show context
ur-whale ◴[05 Mar 23 14:58 UTC] No.35030114[source]▶
Is this the full model or just the weights?

[EDIT]: are there checksums available?

[EDIT2]: MD5 signatures seem to be included for all models in checklist.chk files next to them

And there's also what the author mentions: the magnet file he provides in his README does seed immediately on the download when loaded in a bt app which is usually a good sign that the files are correct.

replies(1): >>35031136 #
rany_ ◴[05 Mar 23 16:38 UTC] No.35031136[source]▶
MD5 signatures don't mean much now that hash collisions could be created instantly on consumer hardware. MD5 is only good for checking for unintentional data corruption.
replies(3): >>35031454 #>>35033009 #>>35033348 #
1. charcircuit ◴[05 Mar 23 20:00 UTC] No.35033348[source]▶
>hash collisions could be created instantly on consumer hardware

Collisions can be created, but MD5 is still preimage resistant. As long as someone with the actual model made the hash and Meta didn't generate colliding models themselves. You can trust it.

replies(1): >>35033747 #
2. ◴[05 Mar 23 20:43 UTC] No.35033747[source]▶