Support finetuning

[fakerybakery]

Hi, can you support finetuning? Llama.cpp supports finetuning. Thank you!

[chrisptang]

looking for the same feature here!

llama.cpp now supports loading with LoRA adapters:
https://github.com/ggerganov/llama.cpp/tree/master/examples/finetune

[abetlen]

Closing this as it's a duplicate of #814

It's on my roadmap but it currently requires some updates to llama.cpp to make this accessible to bindings outside of that project.