Sean Moriarity

Matches the LlamaCPP behavior. I finished the EBNF parser which encodes the grammar in the same way as the implementation from: https://github.com/huggingface/transformers/pull/27557 Unfortunately I think we may have to refactor...

Add zero-shot vision

1

Needs tests

Related to https://github.com/elixir-nx/axon/issues/549 LoRA targets QKV nodes. Right now the only way to do it is by inferring based on layer name. We can provide more information by giving specific...

Return raw errors

2

This is just a proposal, but basically I have a reasonably complex workflow that requires a lot of calls to the OAI API. Because of this, I'm running into rate...

#3 added most vector operations, but does not support partial updates of vectors right now

Right now we allow index creation to `:cuda` (default device) and `{:cuda, ordinal}` specific ordinal. We should support `{:cuda, :all}` which replicates the index to multiple GPUs

Vector transforms in Faiss are pre/post-processing transforms such as PCA or linear transformations. We should add these vector transformations to the API

Selectors in search allow searching for only a subset of IDs in the vector index

The Faiss C++ library raises exceptions for anything that goes wrong, which will crash the VM. We want to catch these exceptions in the NIF and instead return an error...