omarsol's picture
5c4b7bca883ff70dd2c0471b328f95050e328e9ac75b3a3d139d61d93285f7c2
599b0a2 verified
|
raw
history blame
848 Bytes

Configuration

[PeftConfigMixin] is the base configuration class for storing the adapter configuration of a [PeftModel], and [PromptLearningConfig] is the base configuration class for soft prompt methods (p-tuning, prefix tuning, and prompt tuning). These base classes contain methods for saving and loading model configurations from the Hub, specifying the PEFT method to use, type of task to perform, and model configurations like number of layers and number of attention heads.

PeftConfigMixin

[[autodoc]] config.PeftConfigMixin - all

PeftConfig

[[autodoc]] PeftConfig - all

PromptLearningConfig

[[autodoc]] PromptLearningConfig - all