config

mindnlp.peft.config

configs

mindnlp.peft.config.PeftConfig dataclass

Bases: PeftConfigMixin

This is the base configuration class to store the configuration of a [PeftModel].

PARAMETER	DESCRIPTION
peft_type	The type of Peft method to use. TYPE: Union[[`~peft.utils.config.PeftType`], `str`] DEFAULT: None
task_type	The type of task to perform. TYPE: Union[[`~peft.utils.config.TaskType`], `str`] DEFAULT: None
inference_mode	Whether to use the Peft model in inference mode. TYPE: `bool`, defaults to `False` DEFAULT: False

Source code in mindnlp/peft/config.py

@dataclass
class PeftConfig(PeftConfigMixin):
    """
    This is the base configuration class to store the configuration of a [`PeftModel`].

    Args:
        peft_type (Union[[`~peft.utils.config.PeftType`], `str`]): The type of Peft method to use.
        task_type (Union[[`~peft.utils.config.TaskType`], `str`]): The type of task to perform.
        inference_mode (`bool`, defaults to `False`): Whether to use the Peft model in inference mode.
    """

    base_model_name_or_path: str = field(default=None, metadata={"help": "The name of the base model to use."})
    peft_type: Union[str, PeftType] = field(default=None, metadata={"help": "Peft type"})
    task_type: Union[str, TaskType] = field(default=None, metadata={"help": "Task type"})
    inference_mode: bool = field(default=False, metadata={"help": "Whether to use inference mode"})

mindnlp.peft.config.PeftConfigMixin dataclass

This is the base configuration class for PEFT adapter models. It contains all the methods that are common to all PEFT adapter models. The method save_pretrained will save the configuration of your adapter model in a directory. The method from_pretrained will load the configuration of your adapter model from a directory.

PARAMETER	DESCRIPTION
peft_type	The type of Peft method to use. TYPE: Union[[`~peft.utils.config.PeftType`], `str`] DEFAULT: None

Source code in mindnlp/peft/config.py

@dataclass
class PeftConfigMixin():
    r"""
    This is the base configuration class for PEFT adapter models. 
    It contains all the methods that are common to all PEFT adapter models.
    The method `save_pretrained` will save the configuration of your adapter model in a directory.
    The method `from_pretrained` will load the configuration of your adapter model from a directory.

    Args:
        peft_type (Union[[`~peft.utils.config.PeftType`], `str`]): The type of Peft method to use.
    """

    peft_type: Optional[PeftType] = field(default=None, metadata={"help": "The type of PEFT model."})

    @property
    def __dict__(self):
        return asdict(self)

    def to_dict(self):
        """to dict"""
        return self.__dict__

    def save_pretrained(self, save_directory, **kwargs):
        r"""
        This method saves the configuration of your adapter model in a directory.

        Args:
            save_directory (`str`):
                The directory where the configuration will be saved.
            kwargs (additional keyword arguments, *optional*):
                Additional keyword arguments passed along to the 
                [`~transformers.utils.PushToHubMixin.push_to_hub`] method.
        """
        if os.path.isfile(save_directory):
            raise AssertionError(f"Provided path ({save_directory}) should be a directory, not a file")

        os.makedirs(save_directory, exist_ok=True)

        output_dict = asdict(self)
        # converting set type to list
        for key, value in output_dict.items():
            if isinstance(value, set):
                output_dict[key] = list(value)
        output_path = os.path.join(save_directory, CONFIG_NAME)

        # save it
        with open(output_path, "w", encoding='utf-8') as writer:
            writer.write(json.dumps(output_dict, indent=2, sort_keys=True))



    @classmethod
    def from_pretrained(cls, pretrained_model_name_or_path, subfolder=None, **kwargs):
        r"""
        This method loads the configuration of your adapter model from a directory.

        Args:
            pretrained_model_name_or_path (`str`):
                The directory or the Hub repository id where the configuration is saved.
            kwargs (additional keyword arguments, *optional*):
                Additional keyword arguments passed along to the child class initialization.
        """
        path = (
            os.path.join(pretrained_model_name_or_path, subfolder)
            if subfolder is not None
            else pretrained_model_name_or_path
        )
        # read config file
        if os.path.isfile(os.path.join(path, CONFIG_NAME)):
            config_file = os.path.join(path, CONFIG_NAME)
        else:
            raise ValueError(f"Can't find '{CONFIG_NAME}' at '{pretrained_model_name_or_path}'")

        loaded_attributes = cls.from_json_file(config_file)

        config = cls(**kwargs)

        for key, value in loaded_attributes.items():
            if hasattr(config, key):
                setattr(config, key, value)

        return config

    @classmethod
    def from_json_file(cls, path_json_file, **kwargs):
        r"""
        Loads a configuration file from a json file.

        Args:
            path_json_file (`str`):
                The path to the json file.
        """
        with open(path_json_file, "r", encoding='utf-8') as file:
            json_object = json.load(file)

        return json_object


    @property
    def is_prompt_learning(self):
        r"""
        Utility method to check if the configuration is for prompt learning.
        """
        return False

mindnlp.peft.config.PeftConfigMixin.is_prompt_learning property

Utility method to check if the configuration is for prompt learning.

mindnlp.peft.config.PeftConfigMixin.from_json_file(path_json_file, **kwargs) classmethod

Loads a configuration file from a json file.

PARAMETER	DESCRIPTION
path_json_file	The path to the json file. TYPE: `str`

Source code in mindnlp/peft/config.py

@classmethod
def from_json_file(cls, path_json_file, **kwargs):
    r"""
    Loads a configuration file from a json file.

    Args:
        path_json_file (`str`):
            The path to the json file.
    """
    with open(path_json_file, "r", encoding='utf-8') as file:
        json_object = json.load(file)

    return json_object

mindnlp.peft.config.PeftConfigMixin.from_pretrained(pretrained_model_name_or_path, subfolder=None, **kwargs) classmethod

This method loads the configuration of your adapter model from a directory.

PARAMETER	DESCRIPTION
pretrained_model_name_or_path	The directory or the Hub repository id where the configuration is saved. TYPE: `str`
kwargs	Additional keyword arguments passed along to the child class initialization. TYPE: additional keyword arguments, *optional* DEFAULT: {}

Source code in mindnlp/peft/config.py

@classmethod
def from_pretrained(cls, pretrained_model_name_or_path, subfolder=None, **kwargs):
    r"""
    This method loads the configuration of your adapter model from a directory.

    Args:
        pretrained_model_name_or_path (`str`):
            The directory or the Hub repository id where the configuration is saved.
        kwargs (additional keyword arguments, *optional*):
            Additional keyword arguments passed along to the child class initialization.
    """
    path = (
        os.path.join(pretrained_model_name_or_path, subfolder)
        if subfolder is not None
        else pretrained_model_name_or_path
    )
    # read config file
    if os.path.isfile(os.path.join(path, CONFIG_NAME)):
        config_file = os.path.join(path, CONFIG_NAME)
    else:
        raise ValueError(f"Can't find '{CONFIG_NAME}' at '{pretrained_model_name_or_path}'")

    loaded_attributes = cls.from_json_file(config_file)

    config = cls(**kwargs)

    for key, value in loaded_attributes.items():
        if hasattr(config, key):
            setattr(config, key, value)

    return config

mindnlp.peft.config.PeftConfigMixin.save_pretrained(save_directory, **kwargs)

This method saves the configuration of your adapter model in a directory.

PARAMETER	DESCRIPTION
save_directory	The directory where the configuration will be saved. TYPE: `str`
kwargs	Additional keyword arguments passed along to the [~transformers.utils.PushToHubMixin.push_to_hub] method. TYPE: additional keyword arguments, *optional* DEFAULT: {}

Source code in mindnlp/peft/config.py

def save_pretrained(self, save_directory, **kwargs):
    r"""
    This method saves the configuration of your adapter model in a directory.

    Args:
        save_directory (`str`):
            The directory where the configuration will be saved.
        kwargs (additional keyword arguments, *optional*):
            Additional keyword arguments passed along to the 
            [`~transformers.utils.PushToHubMixin.push_to_hub`] method.
    """
    if os.path.isfile(save_directory):
        raise AssertionError(f"Provided path ({save_directory}) should be a directory, not a file")

    os.makedirs(save_directory, exist_ok=True)

    output_dict = asdict(self)
    # converting set type to list
    for key, value in output_dict.items():
        if isinstance(value, set):
            output_dict[key] = list(value)
    output_path = os.path.join(save_directory, CONFIG_NAME)

    # save it
    with open(output_path, "w", encoding='utf-8') as writer:
        writer.write(json.dumps(output_dict, indent=2, sort_keys=True))

mindnlp.peft.config.PeftConfigMixin.to_dict()

to dict

Source code in mindnlp/peft/config.py

def to_dict(self):
    """to dict"""
    return self.__dict__

mindnlp.peft.config.PromptLearningConfig dataclass

Bases: PeftConfig

This is the base configuration class to store the configuration of [PrefixTuning], [PromptEncoder], or [PromptTuning].

PARAMETER	DESCRIPTION
num_virtual_tokens	The number of virtual tokens to use. TYPE: `int` DEFAULT: None
token_dim	The hidden embedding dimension of the base transformer model. TYPE: `int` DEFAULT: None
num_transformer_submodules	The number of transformer submodules in the base transformer model. TYPE: `int` DEFAULT: None
num_attention_heads	The number of attention heads in the base transformer model. TYPE: `int` DEFAULT: None
num_layers	The number of layers in the base transformer model. TYPE: `int` DEFAULT: None

Source code in mindnlp/peft/config.py

@dataclass
class PromptLearningConfig(PeftConfig):
    """
    This is the base configuration class to store the configuration of [`PrefixTuning`], [`PromptEncoder`], or
    [`PromptTuning`].

    Args:
        num_virtual_tokens (`int`): The number of virtual tokens to use.
        token_dim (`int`): The hidden embedding dimension of the base transformer model.
        num_transformer_submodules (`int`): The number of transformer submodules in the base transformer model.
        num_attention_heads (`int`): The number of attention heads in the base transformer model.
        num_layers (`int`): The number of layers in the base transformer model.
    """

    num_virtual_tokens: int = field(default=None, metadata={"help": "Number of virtual tokens"})
    token_dim: int = field(
        default=None, metadata={"help": "The hidden embedding dimension of the base transformer model"}
    )
    num_transformer_submodules: Optional[int] = field(
        default=None, metadata={"help": "Number of transformer submodules"}
    )
    num_attention_heads: Optional[int] = field(default=None, metadata={"help": "Number of attention heads"})
    num_layers: Optional[int] = field(default=None, metadata={"help": "Number of transformer layers"})
    @property
    def is_prompt_learning(self):
        r"""
        Utility method to check if the configuration is for prompt learning.
        """
        return True

mindnlp.peft.config.PromptLearningConfig.is_prompt_learning property

Utility method to check if the configuration is for prompt learning.