mapping

mindnlp.peft.mapping

mappings

mindnlp.peft.mapping.get_peft_config(config_dict)

Returns a Peft config object from a dictionary.

PARAMETER	DESCRIPTION
config_dict	Dictionary containing the configuration parameters. TYPE: `Dict[str, Any]`

Source code in mindnlp/peft/mapping.py

def get_peft_config(config_dict: Dict[str, Any]):
    """
    Returns a Peft config object from a dictionary.

    Args:
        config_dict (`Dict[str, Any]`): Dictionary containing the configuration parameters.
    """
    return PEFT_TYPE_TO_CONFIG_MAPPING[config_dict["peft_type"]](**config_dict)

mindnlp.peft.mapping.get_peft_model(model, peft_config, adapter_name='default')

Returns a Peft model object from a model and a config.

PARAMETER	DESCRIPTION
model	Model to be wrapped. TYPE: [`transformers.PreTrainedModel`]
peft_config	Configuration object containing the parameters of the Peft model. TYPE: [`PeftConfig`]

Source code in mindnlp/peft/mapping.py

def get_peft_model(model: nn.Module, peft_config: PeftConfig, adapter_name: str = "default") -> PeftModel:
    """
    Returns a Peft model object from a model and a config.

    Args:
        model ([`transformers.PreTrainedModel`]): Model to be wrapped.
        peft_config ([`PeftConfig`]): Configuration object containing the parameters of the Peft model.
    """
    model_config = getattr(model, "config", {"model_type": "custom"})
    if hasattr(model_config, "to_dict"):
        model_config = model_config.to_dict()
    peft_config.base_model_name_or_path = model.__dict__.get("name_or_path", None)

    # no specific task_type and is not prompt_learning
    if peft_config.task_type not in MODEL_TYPE_TO_PEFT_MODEL_MAPPING.keys() and not peft_config.is_prompt_learning:
        return PeftModel(model, peft_config, adapter_name=adapter_name)

    # TODO: prompt learning
    # if peft_config.is_prompt_learning:
    #     # peft_config = _prepare_prompt_learning_config(peft_config, model_config)
    return MODEL_TYPE_TO_PEFT_MODEL_MAPPING[peft_config.task_type](model, peft_config, adapter_name=adapter_name)

mindnlp.peft.mapping.inject_adapter_in_model(peft_config, model, adapter_name='default')

A simple API to create and inject adapter in-place into a model. Currently the API does not support prompt learning methods and adaption prompt. Make sure to have the correct target_names set in the peft_config object. The API calls get_peft_model under the hood but would be restricted only to non-prompt learning methods.

PARAMETER	DESCRIPTION
peft_config	Configuration object containing the parameters of the Peft model. TYPE: `PeftConfig`
model	The input model where the adapter will be injected. TYPE: `nn.Module`
adapter_name	The name of the adapter to be injected, if not provided, the default adapter name is used ("default"). TYPE: `str`, `optional`, defaults to `"default"` DEFAULT: 'default'

Source code in mindnlp/peft/mapping.py

def inject_adapter_in_model(
    peft_config: PeftConfig, model: nn.Module, adapter_name: str = "default"
) -> nn.Module:
    r"""
    A simple API to create and inject adapter in-place into a model. Currently the API does not support prompt learning
    methods and adaption prompt. Make sure to have the correct `target_names` set in the `peft_config` object. The API
    calls `get_peft_model` under the hood but would be restricted only to non-prompt learning methods.

    Args:
        peft_config (`PeftConfig`):
            Configuration object containing the parameters of the Peft model.
        model (`nn.Module`):
            The input model where the adapter will be injected.
        adapter_name (`str`, `optional`, defaults to `"default"`):
            The name of the adapter to be injected, if not provided, the default adapter name is used ("default").
    """
    if peft_config.is_prompt_learning or peft_config.is_adaption_prompt:
        raise ValueError("`create_and_replace` does not support prompt learning and adaption prompt yet.")

    if peft_config.peft_type not in PEFT_TYPE_TO_TUNER_MAPPING.keys():
        raise ValueError(
            f"`inject_adapter_in_model` does not support {peft_config.peft_type} yet. Please use `get_peft_model`."
        )

    tuner_cls = PEFT_TYPE_TO_TUNER_MAPPING[peft_config.peft_type]

    # By instantiating a peft model we are injecting randomly initialized LoRA layers into the model's cells.
    peft_model = tuner_cls(model, peft_config, adapter_name=adapter_name)

    return peft_model.model