News for hackers. Stuff that matters.

A comment on Launch HN: Outerport (YC S24) – Instant hot-swapping for AI model weights.

Re:Launch HN: Outerport (YC S24) – Instant hot-swapping for ... by bravura on Wednesday August 21, 2024 @06:17PM (#41314748)
Do all variations of the model need to have the same architecture?

Or can they be different types of models with different number of layers, etc?

[ Parent | on HN ]
Re:Launch HN: Outerport (YC S24) – Instant hot-swapping for ... by tovacinni on Wednesday August 21, 2024 @06:24PM (#41314794)
Variants do not have to be the same architecture- the demo (https://hotswap.outerport.com/) runs on a couple of different open source architectures.

That being said, there is some smart caching / hashing on layers such that if you do have models that are similar (i.e. a fine-tuned model where only some layers are fine-tuned), it'll minimize storage and transfer by reusing those weights.

[ Parent | on HN ]