Dynamic LoRA adapters

[micmcg]

Do you have plans to add support for dynamically loading/unloading LoRA adapters as in the upstream vLLM?
https://docs.vllm.ai/en/stable/features/lora/#dynamically-serving-lora-adapters

Thanks!