PEFT LORA / QLORA

[louisoutin]

Hello, Are you planning to add support for parameter efficient finetuning methods? Also does it support doing inference using those adapters models to optimize VRAM? Thanks