196

16412 readers

1168 users here now

Be sure to follow the rule before you head out.

Rule: You must post before you leave.

^other^ ^rules^

founded 1 year ago

MODERATORS

[email protected]

657

The Rule (lemmy.ml)

submitted 3 months ago by [email protected] to c/[email protected]

63 comments fedilink hide all child comments

you are viewing a single comment's thread
view the rest of the comments

[–] [email protected] 9 points 3 months ago (1 children)

There's quantization which basically compresses the model to use a smaller data type for each weight. Reduces memory requirements by half or even more.

There's also airllm which loads a part of the model into RAM, runs those calculations, unloads that part, loads the next part, etc... It's a nice option but the performance of all that loading/unloading is never going to be great, especially on a huge model like llama 405b

Then there are some neat projects to distribute models across multiple computers like exo and petals. They're more targeted at a p2p-style random collection of computers. I've run petals in a small cluster and it works reasonably well.

[–] [email protected] 1 points 3 months ago

Yes, but 200 gb is probably already with 4 bit quantization, the weights in fp16 would be more like 800 gb IDK if its even possible to quantize more, if it is, you're probably better of going with a smaller model anyways