transformers/docs/source/en/main_classes/peft.md

PEFT

The [~integrations.PeftAdapterMixin] provides functions from the PEFT library for managing adapters with Transformers. This mixin currently supports LoRA, IA3, and AdaLora. Prefix tuning methods (prompt tuning, prompt learning) aren't supported because they can't be injected into a torch module.

[[autodoc]] integrations.PeftAdapterMixin - load_adapter - add_adapter - set_adapter - disable_adapters - enable_adapters - active_adapters - get_adapter_state_dict