PaddleX/paddlex/inference/models/common/tokenizer/tokenizer_utils_base.py

# Copyright (c) 2024 PaddlePaddle Authors. All Rights Reserved.
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
#    http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.

import copy
import inspect
import io
import json
import os
import warnings
from collections import UserDict
from dataclasses import dataclass, field
from enum import Enum
from typing import (
    Any,
    Dict,
    List,
    Literal,
    NamedTuple,
    Optional,
    Sequence,
    Tuple,
    Union,
)

import numpy as np

from .....utils import logging

__all__ = [
    "AddedToken",
    "FastEncoding",
    "ExplicitEnum",
    "PaddingStrategy",
    "TensorType",
    "TruncationStrategy",
    "CharSpan",
    "TokenSpan",
    "BatchEncoding",
    "SpecialTokensMixin",
    "PretrainedTokenizerBase",
]

TOKENIZER_CONFIG_NAME = "tokenizer_config.json"
CHAT_TEMPLATE_CONFIG_NAME = "chat_template.json"

VERY_LARGE_INTEGER = int(
    1e30
)  # This is used to set the max input length for a model with infinite size input
LARGE_INTEGER = int(
    1e20
)  # This is used when we need something big but slightly smaller than VERY_LARGE_INTEGER

# Define type aliases and NamedTuples
TextInput = str
PreTokenizedInput = List[str]
EncodedInput = List[int]
TextInputPair = Tuple[str, str]
PreTokenizedInputPair = Tuple[List[str], List[str]]
EncodedInputPair = Tuple[List[int], List[int]]

# Slow tokenizers used to be saved in three separated files
SPECIAL_TOKENS_MAP_FILE = "special_tokens_map.json"
ADDED_TOKENS_FILE = "added_tokens.json"
TOKENIZER_CONFIG_FILE = "tokenizer_config.json"


@dataclass(frozen=True, eq=True)
class AddedToken:
    """
    AddedToken represents a token to be added to a Tokenizer An AddedToken can have special options defining the
    way it should behave.
    """

    content: str = field(default_factory=str)
    single_word: bool = False
    lstrip: bool = False
    rstrip: bool = False
    normalized: bool = True
    special: bool = True

    def __getstate__(self):
        return self.__dict__

    def __str__(self):
        return self.content


@dataclass
class FastEncoding:
    """This is dummy class reserved for fast tokenizer"""


class ExplicitEnum(Enum):
    """
    Enum with more explicit error message for missing values.
    """

    @classmethod
    def _missing_(cls, value):
        raise ValueError(
            f"{value} is not a valid {cls.__name__}, please select one of {list(cls._value2member_map_.keys())}"
        )


class PaddingStrategy(ExplicitEnum):
    """
    Possible values for the `padding` argument in [`PretrainedTokenizerBase.__call__`]. Useful for tab-completion in an
    IDE.
    """

    LONGEST = "longest"
    MAX_LENGTH = "max_length"
    DO_NOT_PAD = "do_not_pad"


class TensorType(ExplicitEnum):
    """
    Possible values for the `return_tensors` argument in [`PretrainedTokenizerBase.__call__`]. Useful for
    tab-completion in an IDE.
    """

    PADDLE = "pd"
    NUMPY = "np"


def to_py_obj(obj):
    """
    Convert a Paddle tensor, Numpy array or python list to a python list.
    """
    import paddle

    if isinstance(obj, (dict, UserDict)):
        return {k: to_py_obj(v) for k, v in obj.items()}
    elif isinstance(obj, (list, tuple)):
        return [to_py_obj(o) for o in obj]
    elif isinstance(obj, paddle.Tensor):
        return obj.numpy().tolist()
    elif isinstance(obj, (np.ndarray, np.number)):  # tolist also works on 0d np arrays
        return obj.tolist()
    else:
        return obj


def _is_numpy(x):
    return isinstance(x, np.ndarray)


class TruncationStrategy(ExplicitEnum):
    """
    Possible values for the `truncation` argument in [`PretrainedTokenizerBase.__call__`]. Useful for tab-completion in
    an IDE.
    """

    ONLY_FIRST = "only_first"
    ONLY_SECOND = "only_second"
    LONGEST_FIRST = "longest_first"
    DO_NOT_TRUNCATE = "do_not_truncate"


class CharSpan(NamedTuple):
    """
    Character span in the original string.

    Args:
        start (`int`): Index of the first character in the original string.
        end (`int`): Index of the character following the last character in the original string.
    """

    start: int
    end: int


class TokenSpan(NamedTuple):
    """
    Token span in an encoded string (list of tokens).

    Args:
        start (`int`): Index of the first token in the span.
        end (`int`): Index of the token following the last token in the span.
    """

    start: int
    end: int


class BatchEncoding(UserDict):
    """
    Holds the output of the [`PretrainedTokenizerBase.__call__`],
    [`PretrainedTokenizerBase.encode_plus`] and
    [`PretrainedTokenizerBase.batch_encode_plus`] methods (tokens, attention_masks, etc).

    This class is derived from a python dictionary and can be used as a dictionary. In addition, this class exposes
    utility methods to map from word/character space to token space.

    Args:
        data (`dict`):
            Dictionary of lists/arrays/tensors returned by the `__call__`/`encode`/`batch_encode` methods
            ('input_ids', 'attention_mask', etc.).
        tensor_type (`Union[None, str, TensorType]`, *optional*):
            You can give a tensor_type here to convert the lists of integers in Paddle/Numpy Tensors at
            initialization.
        prepend_batch_axis (`bool`, *optional*, defaults to `False`):
            Whether or not to add a batch axis when converting to tensors (see `tensor_type` above).
    """

    def __init__(
        self,
        data: Optional[Dict[str, Any]] = None,
        encoding: Optional[Union[FastEncoding, Sequence[FastEncoding]]] = None,
        tensor_type: Union[None, str] = None,
        prepend_batch_axis: bool = False,
        n_sequences: Optional[int] = None,
    ):
        super().__init__(data)

        if isinstance(encoding, FastEncoding):
            encoding = [encoding]

        self._encodings = encoding

        if n_sequences is None and encoding is not None and len(encoding):
            n_sequences = encoding[0].n_sequences

        self._n_sequences = n_sequences

        self.convert_to_tensors(
            tensor_type=tensor_type, prepend_batch_axis=prepend_batch_axis
        )

    @property
    def n_sequences(self) -> Optional[int]:
        """
        `Optional[int]`: The number of sequences used to generate each sample from the batch encoded in this
        [`BatchEncoding`]. Currently can be one of `None` (unknown), `1` (a single sentence) or `2` (a pair of
        sentences)
        """
        return self._n_sequences

    @property
    def is_fast(self) -> bool:
        """
        `bool`: Indicate whether this [`BatchEncoding`] was generated from the result of a [`PretrainedFastTokenizer`]
        or not.
        """
        return self._encodings is not None

    def __getitem__(self, item: Union[int, str]) -> Union[Any, FastEncoding]:
        """
        If the key is a string, returns the value of the dict associated to `key` ('input_ids', 'attention_mask',
        etc.).

        If the key is an integer, get the `Encoding` for batch item with index `key`.
        """
        if isinstance(item, str):
            return self.data[item]
        elif self._encodings is not None:
            return self._encodings[item]
        else:
            raise KeyError(
                "Indexing with integers is not available when using tokenizer.__call__()"
                " with return_dict=True. Please set return_dict to False to use integer indexing."
            )

    def __getattr__(self, item: str):
        try:
            return self.data[item]
        except KeyError:
            raise AttributeError

    def __getstate__(self):
        return {"data": self.data, "encodings": self._encodings}

    def __setstate__(self, state):
        if "data" in state:
            self.data = state["data"]

        if "encodings" in state:
            self._encodings = state["encodings"]

    def keys(self):
        return self.data.keys()

    def values(self):
        return self.data.values()

    def items(self):
        return self.data.items()

    @property
    def encodings(self) -> Optional[List[FastEncoding]]:
        """
        `Optional[List[FastEncoding]]`: The list all encodings from the tokenization process. Returns `None` if
        the input was tokenized through Python (i.e., not a fast) tokenizer.
        """
        return self._encodings

    def tokens(self, batch_index: int = 0) -> List[str]:
        """
        Return the list of tokens (sub-parts of the input strings after word/subword splitting and before conversion to
        integer indices) at a given batch index (only works for the output of a fast tokenizer).

        Args:
            batch_index (`int`, *optional*, defaults to 0): The index to access in the batch.

        Returns:
            `List[str]`: The list of tokens at that index.
        """
        if not self._encodings:
            raise ValueError(
                "tokens() is not available when using Python-based tokenizers"
            )
        return self._encodings[batch_index].tokens

    def sequence_ids(self, batch_index: int = 0) -> List[Optional[int]]:
        """
        Return a list mapping the tokens to the id of their original sentences:

            - `None` for special tokens added around or between sequences,
            - `0` for tokens corresponding to words in the first sequence,
            - `1` for tokens corresponding to words in the second sequence when a pair of sequences was jointly
              encoded.

        Args:
            batch_index (`int`, *optional*, defaults to 0): The index to access in the batch.

        Returns:
            `List[Optional[int]]`: A list indicating the sequence id corresponding to each token. Special tokens added
            by the tokenizer are mapped to `None` and other tokens are mapped to the index of their corresponding
            sequence.
        """
        if not self._encodings:
            raise ValueError(
                "sequence_ids() is not available when using Python-based tokenizers"
            )
        return self._encodings[batch_index].sequence_ids

    def words(self, batch_index: int = 0) -> List[Optional[int]]:
        """
        Return a list mapping the tokens to their actual word in the initial sentence for a fast tokenizer.

        Args:
            batch_index (`int`, *optional*, defaults to 0): The index to access in the batch.

        Returns:
            `List[Optional[int]]`: A list indicating the word corresponding to each token. Special tokens added by the
            tokenizer are mapped to `None` and other tokens are mapped to the index of their corresponding word
            (several tokens will be mapped to the same word index if they are parts of that word).
        """
        if not self._encodings:
            raise ValueError(
                "words() is not available when using Python-based tokenizers"
            )
        warnings.warn(
            "`BatchEncoding.words()` property is deprecated and should be replaced with the identical, "
            "but more self-explanatory `BatchEncoding.word_ids()` property.",
            FutureWarning,
        )
        return self.word_ids(batch_index)

    def word_ids(self, batch_index: int = 0) -> List[Optional[int]]:
        """
        Return a list mapping the tokens to their actual word in the initial sentence for a fast tokenizer.

        Args:
            batch_index (`int`, *optional*, defaults to 0): The index to access in the batch.

        Returns:
            `List[Optional[int]]`: A list indicating the word corresponding to each token. Special tokens added by the
            tokenizer are mapped to `None` and other tokens are mapped to the index of their corresponding word
            (several tokens will be mapped to the same word index if they are parts of that word).
        """
        if not self._encodings:
            raise ValueError(
                "word_ids() is not available when using Python-based tokenizers"
            )
        return self._encodings[batch_index].word_ids

    def token_to_sequence(
        self, batch_or_token_index: int, token_index: Optional[int] = None
    ) -> int:
        """
        Get the index of the sequence represented by the given token. In the general use case, this method returns `0`
        for a single sequence or the first sequence of a pair, and `1` for the second sequence of a pair

        Can be called as:

        - `self.token_to_sequence(token_index)` if batch size is 1
        - `self.token_to_sequence(batch_index, token_index)` if batch size is greater than 1

        This method is particularly suited when the input sequences are provided as pre-tokenized sequences (i.e.,
        words are defined by the user). In this case it allows to easily associate encoded tokens with provided
        tokenized words.

        Args:
            batch_or_token_index (`int`):
                Index of the sequence in the batch. If the batch only comprises one sequence, this can be the index of
                the token in the sequence.
            token_index (`int`, *optional*):
                If a batch index is provided in *batch_or_token_index*, this can be the index of the token in the
                sequence.

        Returns:
            `int`: Index of the word in the input sequence.
        """

        if not self._encodings:
            raise ValueError(
                "token_to_sequence() is not available when using Python based tokenizers"
            )
        if token_index is not None:
            batch_index = batch_or_token_index
        else:
            batch_index = 0
            token_index = batch_or_token_index
        if batch_index < 0:
            batch_index = self._batch_size + batch_index
        if token_index < 0:
            token_index = self._seq_len + token_index
        return self._encodings[batch_index].token_to_sequence(token_index)

    def token_to_word(
        self, batch_or_token_index: int, token_index: Optional[int] = None
    ) -> int:
        """
        Get the index of the word corresponding (i.e. comprising) to an encoded token in a sequence of the batch.

        Can be called as:

        - `self.token_to_word(token_index)` if batch size is 1
        - `self.token_to_word(batch_index, token_index)` if batch size is greater than 1

        This method is particularly suited when the input sequences are provided as pre-tokenized sequences (i.e.,
        words are defined by the user). In this case it allows to easily associate encoded tokens with provided
        tokenized words.

        Args:
            batch_or_token_index (`int`):
                Index of the sequence in the batch. If the batch only comprise one sequence, this can be the index of
                the token in the sequence.
            token_index (`int`, *optional*):
                If a batch index is provided in *batch_or_token_index*, this can be the index of the token in the
                sequence.

        Returns:
            `int`: Index of the word in the input sequence.
        """

        if not self._encodings:
            raise ValueError(
                "token_to_word() is not available when using Python based tokenizers"
            )
        if token_index is not None:
            batch_index = batch_or_token_index
        else:
            batch_index = 0
            token_index = batch_or_token_index
        if batch_index < 0:
            batch_index = self._batch_size + batch_index
        if token_index < 0:
            token_index = self._seq_len + token_index
        return self._encodings[batch_index].token_to_word(token_index)

    def word_to_tokens(
        self,
        batch_or_word_index: int,
        word_index: Optional[int] = None,
        sequence_index: int = 0,
    ) -> Optional[TokenSpan]:
        """
        Get the encoded token span corresponding to a word in a sequence of the batch.

        Token spans are returned as a [`TokenSpan`] with:

        - **start** -- Index of the first token.
        - **end** -- Index of the token following the last token.

        Can be called as:

        - `self.word_to_tokens(word_index, sequence_index: int = 0)` if batch size is 1
        - `self.word_to_tokens(batch_index, word_index, sequence_index: int = 0)` if batch size is greater or equal to
          1

        This method is particularly suited when the input sequences are provided as pre-tokenized sequences (i.e. words
        are defined by the user). In this case it allows to easily associate encoded tokens with provided tokenized
        words.

        Args:
            batch_or_word_index (`int`):
                Index of the sequence in the batch. If the batch only comprises one sequence, this can be the index of
                the word in the sequence.
            word_index (`int`, *optional*):
                If a batch index is provided in *batch_or_token_index*, this can be the index of the word in the
                sequence.
            sequence_index (`int`, *optional*, defaults to 0):
                If pair of sequences are encoded in the batch this can be used to specify which sequence in the pair (0
                or 1) the provided word index belongs to.

        Returns:
            Optional [`TokenSpan`] Span of tokens in the encoded sequence. Returns `None` if
            no tokens correspond to the word.
        """

        if not self._encodings:
            raise ValueError(
                "word_to_tokens() is not available when using Python based tokenizers"
            )
        if word_index is not None:
            batch_index = batch_or_word_index
        else:
            batch_index = 0
            word_index = batch_or_word_index
        if batch_index < 0:
            batch_index = self._batch_size + batch_index
        if word_index < 0:
            word_index = self._seq_len + word_index
        span = self._encodings[batch_index].word_to_tokens(word_index, sequence_index)
        return TokenSpan(*span) if span is not None else None

    def token_to_chars(
        self, batch_or_token_index: int, token_index: Optional[int] = None
    ) -> CharSpan:
        """
        Get the character span corresponding to an encoded token in a sequence of the batch.

        Character spans are returned as a [`CharSpan`] with:

        - **start** -- Index of the first character in the original string associated to the token.
        - **end** -- Index of the character following the last character in the original string associated to the
          token.

        Can be called as:

        - `self.token_to_chars(token_index)` if batch size is 1
        - `self.token_to_chars(batch_index, token_index)` if batch size is greater or equal to 1

        Args:
            batch_or_token_index (`int`):
                Index of the sequence in the batch. If the batch only comprise one sequence, this can be the index of
                the token in the sequence.
            token_index (`int`, *optional*):
                If a batch index is provided in *batch_or_token_index*, this can be the index of the token or tokens in
                the sequence.

        Returns:
            [`CharSpan`]: Span of characters in the original string.
        """

        if not self._encodings:
            raise ValueError(
                "token_to_chars() is not available when using Python based tokenizers"
            )
        if token_index is not None:
            batch_index = batch_or_token_index
        else:
            batch_index = 0
            token_index = batch_or_token_index
        return CharSpan(*(self._encodings[batch_index].token_to_chars(token_index)))

    def char_to_token(
        self,
        batch_or_char_index: int,
        char_index: Optional[int] = None,
        sequence_index: int = 0,
    ) -> int:
        """
        Get the index of the token in the encoded output comprising a character in the original string for a sequence
        of the batch.

        Can be called as:

        - `self.char_to_token(char_index)` if batch size is 1
        - `self.char_to_token(batch_index, char_index)` if batch size is greater or equal to 1

        This method is particularly suited when the input sequences are provided as pre-tokenized sequences (i.e. words
        are defined by the user). In this case it allows to easily associate encoded tokens with provided tokenized
        words.

        Args:
            batch_or_char_index (`int`):
                Index of the sequence in the batch. If the batch only comprise one sequence, this can be the index of
                the word in the sequence
            char_index (`int`, *optional*):
                If a batch index is provided in *batch_or_token_index*, this can be the index of the word in the
                sequence.
            sequence_index (`int`, *optional*, defaults to 0):
                If pair of sequences are encoded in the batch this can be used to specify which sequence in the pair (0
                or 1) the provided character index belongs to.


        Returns:
            `int`: Index of the token.
        """

        if not self._encodings:
            raise ValueError(
                "char_to_token() is not available when using Python based tokenizers"
            )
        if char_index is not None:
            batch_index = batch_or_char_index
        else:
            batch_index = 0
            char_index = batch_or_char_index
        return self._encodings[batch_index].char_to_token(char_index, sequence_index)

    def word_to_chars(
        self,
        batch_or_word_index: int,
        word_index: Optional[int] = None,
        sequence_index: int = 0,
    ) -> CharSpan:
        """
        Get the character span in the original string corresponding to given word in a sequence of the batch.

        Character spans are returned as a CharSpan NamedTuple with:

        - start: index of the first character in the original string
        - end: index of the character following the last character in the original string

        Can be called as:

        - `self.word_to_chars(word_index)` if batch size is 1
        - `self.word_to_chars(batch_index, word_index)` if batch size is greater or equal to 1

        Args:
            batch_or_word_index (`int`):
                Index of the sequence in the batch. If the batch only comprise one sequence, this can be the index of
                the word in the sequence
            word_index (`int`, *optional*):
                If a batch index is provided in *batch_or_token_index*, this can be the index of the word in the
                sequence.
            sequence_index (`int`, *optional*, defaults to 0):
                If pair of sequences are encoded in the batch this can be used to specify which sequence in the pair (0
                or 1) the provided word index belongs to.

        Returns:
            `CharSpan` or `List[CharSpan]`: Span(s) of the associated character or characters in the string. CharSpan
            are NamedTuple with:

                - start: index of the first character associated to the token in the original string
                - end: index of the character following the last character associated to the token in the original
                  string
        """

        if not self._encodings:
            raise ValueError(
                "word_to_chars() is not available when using Python based tokenizers"
            )
        if word_index is not None:
            batch_index = batch_or_word_index
        else:
            batch_index = 0
            word_index = batch_or_word_index
        return CharSpan(
            *(self._encodings[batch_index].word_to_chars(word_index, sequence_index))
        )

    def char_to_word(
        self,
        batch_or_char_index: int,
        char_index: Optional[int] = None,
        sequence_index: int = 0,
    ) -> int:
        """
        Get the word in the original string corresponding to a character in the original string of a sequence of the
        batch.

        Can be called as:

        - `self.char_to_word(char_index)` if batch size is 1
        - `self.char_to_word(batch_index, char_index)` if batch size is greater than 1

        This method is particularly suited when the input sequences are provided as pre-tokenized sequences (i.e. words
        are defined by the user). In this case it allows to easily associate encoded tokens with provided tokenized
        words.

        Args:
            batch_or_char_index (`int`):
                Index of the sequence in the batch. If the batch only comprise one sequence, this can be the index of
                the character in the original string.
            char_index (`int`, *optional*):
                If a batch index is provided in *batch_or_token_index*, this can be the index of the character in the
                original string.
            sequence_index (`int`, *optional*, defaults to 0):
                If pair of sequences are encoded in the batch this can be used to specify which sequence in the pair (0
                or 1) the provided character index belongs to.


        Returns:
            `int` or `List[int]`: Index or indices of the associated encoded token(s).
        """

        if not self._encodings:
            raise ValueError(
                "char_to_word() is not available when using Python based tokenizers"
            )
        if char_index is not None:
            batch_index = batch_or_char_index
        else:
            batch_index = 0
            char_index = batch_or_char_index
        return self._encodings[batch_index].char_to_word(char_index, sequence_index)

    def convert_to_tensors(
        self,
        tensor_type: Optional[Union[str, TensorType]] = None,
        prepend_batch_axis: bool = False,
    ):
        """
        Convert the inner content to tensors.

        Args:
            tensor_type (`str` or [`TensorType`], *optional*):
                The type of tensors to use. If `str`, should be one of the values of the enum [`TensorType`]. If
                `None`, no modification is done.
            prepend_batch_axis (`int`, *optional*, defaults to `False`):
                Whether or not to add the batch dimension during the conversion.
        """
        import paddle

        if tensor_type is None:
            return self

        # Convert to TensorType
        if not isinstance(tensor_type, TensorType):
            tensor_type = TensorType(tensor_type)
        # Get a function reference for the correct framework
        if tensor_type == TensorType.PADDLE:
            as_tensor = paddle.to_tensor
            is_tensor = paddle.is_tensor
        else:
            as_tensor = np.asarray
            is_tensor = _is_numpy

        # Do the tensor conversion in batch
        for key, value in self.items():
            try:
                if prepend_batch_axis:
                    value = [value]

                if not is_tensor(value):
                    tensor = as_tensor(value)

                    self[key] = tensor
            except:  # noqa E722
                if key == "overflowing_tokens":
                    raise ValueError(
                        "Unable to create tensor returning overflowing tokens of different lengths. "
                        "Please see if a fast version of this tokenizer is available to have this feature available."
                    )
                raise ValueError(
                    "Unable to create tensor, you should probably activate truncation and/or padding "
                    "with 'padding=True' 'truncation=True' to have batched tensors with the same length."
                )

        return self


class SpecialTokensMixin:
    """
    A mixin derived by [`PretrainedTokenizer`] to handle specific behaviors related to
    special tokens. In particular, this class hold the attributes which can be used to directly access these special
    tokens in a model-independent manner and allow to set and update the special tokens.

    Args:
        bos_token (`str` or `AddedToken`, *optional*):
            A special token representing the beginning of a sentence.
        eos_token (`str` or `AddedToken`, *optional*):
            A special token representing the end of a sentence.
        unk_token (`str` or `AddedToken`, *optional*):
            A special token representing an out-of-vocabulary token.
        sep_token (`str` or `AddedToken`, *optional*):
            A special token separating two different sentences in the same input (used by BERT for instance).
        pad_token (`str` or `AddedToken`, *optional*):
            A special token used to make arrays of tokens the same size for batching purpose. Will then be ignored by
            attention mechanisms or loss computation.
        cls_token (`str` or `AddedToken`, *optional*):
            A special token representing the class of the input (used by BERT for instance).
        mask_token (`str` or `AddedToken`, *optional*):
            A special token representing a masked token (used by masked-language modeling pretraining objectives, like
            BERT).
        additional_special_tokens (tuple or list of `str` or `AddedToken`, *optional*):
            A tuple or a list of additional special tokens.
    """

    SPECIAL_TOKENS_ATTRIBUTES = [
        "bos_token",
        "eos_token",
        "unk_token",
        "sep_token",
        "pad_token",
        "cls_token",
        "mask_token",
        "additional_special_tokens",
    ]

    def __init__(self, verbose=True, **kwargs):
        # note(guosheng): Since `__init__` might be called multiple times which
        # is hooked before `PretrainedTokenizer` init, we do not set to None as
        # HF to avoid unintentional overriding.
        self._bos_token = getattr(self, "_bos_token", None)
        self._eos_token = getattr(self, "_eos_token", None)
        self._unk_token = getattr(self, "_unk_token", None)
        self._sep_token = getattr(self, "_sep_token", None)
        self._pad_token = getattr(self, "_pad_token", None)
        self._cls_token = getattr(self, "_cls_token", None)
        self._mask_token = getattr(self, "_mask_token", None)
        self._pad_token_type_id = getattr(self, "_pad_token_type_id", 0)
        self._additional_special_tokens = getattr(
            self, "_additional_special_tokens", []
        )
        self.verbose = verbose

        # We directly set the hidden value to allow initialization with special tokens
        # which are not yet in the vocabulary. Necessary for serialization/de-serialization
        # TODO clean this up at some point (probably by switching to fast tokenizers)
        for key, value in kwargs.items():
            if value is None:
                continue
            if key in self.SPECIAL_TOKENS_ATTRIBUTES:
                if key == "additional_special_tokens":
                    assert isinstance(
                        value, (list, tuple)
                    ), f"Value {value} is not a list or tuple"
                    assert all(
                        isinstance(t, (str, AddedToken)) for t in value
                    ), "One of the tokens is not a string or an AddedToken"
                    setattr(self, key, value)
                elif isinstance(value, (str, AddedToken)):
                    setattr(self, key, value)
                else:
                    raise TypeError(
                        f"special token {key} has to be either str or AddedToken but got: {type(value)}"
                    )

    def sanitize_special_tokens(self) -> int:
        """
        Make sure that all the special tokens attributes of the tokenizer (`tokenizer.mask_token`,
        `tokenizer.cls_token`, etc.) are in the vocabulary.

        Add the missing ones to the vocabulary if needed.

        Return:
            `int`: The number of tokens added in the vocabulary during the operation.
        """
        return self.add_tokens(self.all_special_tokens_extended, special_tokens=True)

    def add_special_tokens(
        self,
        special_tokens_dict: Dict[str, Union[str, AddedToken]],
        replace_additional_special_tokens=True,
    ) -> int:
        """
        Add a dictionary of special tokens (eos, pad, cls, etc.) to the encoder and link them to class attributes. If
        special tokens are NOT in the vocabulary, they are added to it (indexed starting from the last index of the
        current vocabulary).

        When adding new tokens to the vocabulary, you should make sure to also resize the token embedding matrix of the
        model so that its embedding matrix matches the tokenizer.

        In order to do that, please use the [`~PreTrainedModel.resize_token_embeddings`] method.

        Using `add_special_tokens` will ensure your special tokens can be used in several ways:

        - Special tokens are carefully handled by the tokenizer (they are never split).
        - You can easily refer to special tokens using tokenizer class attributes like `tokenizer.cls_token`. This
          makes it easy to develop model-agnostic training and fine-tuning scripts.

        When possible, special tokens are already registered for provided pretrained models (for instance
        [`BertTokenizer`] `cls_token` is already registered to be :obj*'[CLS]'* and XLM's one is also registered to be
        `'</s>'`).

        Args:
            special_tokens_dict (dictionary *str* to *str* or `AddedToken`):
                Keys should be in the list of predefined special attributes: [`bos_token`, `eos_token`, `unk_token`,
                `sep_token`, `pad_token`, `cls_token`, `mask_token`, `additional_special_tokens`].

                Tokens are only added if they are not already in the vocabulary (tested by checking if the tokenizer
                assign the index of the `unk_token` to them).
            replace_additional_special_tokens (`bool`, *optional*,, defaults to `True`):
                If `True`, the existing list of additional special tokens will be replaced by the list provided in
                `special_tokens_dict`. Otherwise, `self._additional_special_tokens` is just extended. In the former
                case, the tokens will NOT be removed from the tokenizer's full vocabulary - they are only being flagged
                as non-special tokens. Remember, this only affects which tokens are skipped during decoding, not the
                `added_tokens_encoder` and `added_tokens_decoder`. This means that the previous
                `additional_special_tokens` are still added tokens, and will not be split by the model.

        Returns:
            `int`: Number of tokens added to the vocabulary.

        Examples:

        ```python
        # Let's see how to add a new classification token to GPT-2
        tokenizer = GPT2Tokenizer.from_pretrained("gpt2")
        model = GPT2Model.from_pretrained("gpt2")

        special_tokens_dict = {"cls_token": "<CLS>"}

        num_added_toks = tokenizer.add_special_tokens(special_tokens_dict)
        print("We have added", num_added_toks, "tokens")
        # Notice: resize_token_embeddings expect to receive the full size of the new vocabulary, i.e., the length of the tokenizer.
        model.resize_token_embeddings(len(tokenizer))

        assert tokenizer.cls_token == "<CLS>"
        ```"""
        if not special_tokens_dict:
            return 0

        added_tokens = []
        for key, value in special_tokens_dict.items():
            assert (
                key in self.SPECIAL_TOKENS_ATTRIBUTES
            ), f"Key {key} is not a special token"

            if self.verbose:
                logging.info(f"Assigning {value} to the {key} key of the tokenizer")

            if key == "additional_special_tokens":
                assert isinstance(value, (list, tuple)) and all(
                    isinstance(t, (str, AddedToken)) for t in value
                ), f"Tokens {value} for key {key} should all be str or AddedToken instances"

                to_add = []
                for token in value:
                    if (
                        not replace_additional_special_tokens
                        and str(token) in self.additional_special_tokens
                    ):
                        continue
                    to_add.append(token)
                if replace_additional_special_tokens and len(to_add) > 0:
                    setattr(self, key, list(to_add))
                else:
                    self._additional_special_tokens.extend(to_add)
                added_tokens += to_add

            else:
                if not isinstance(value, (str, AddedToken)):
                    raise ValueError(
                        f"Token {value} for key {key} should be a str or an AddedToken instance"
                    )
                setattr(self, key, value)
                if value not in added_tokens:
                    added_tokens.append(value)

        # if we are adding tokens that were not part of the vocab, we ought to add them
        added_tokens = self.add_tokens(added_tokens, special_tokens=True)
        return added_tokens

    def add_tokens(
        self,
        new_tokens: Union[str, AddedToken, List[Union[str, AddedToken]]],
        special_tokens: bool = False,
    ) -> int:
        """
        Add a list of new tokens to the tokenizer class. If the new tokens are not in the vocabulary, they are added to
        it with indices starting from length of the current vocabulary.

        Note,None When adding new tokens to the vocabulary, you should make sure to also resize the token embedding
        matrix of the model so that its embedding matrix matches the tokenizer.

        In order to do that, please use the [`~PreTrainedModel.resize_token_embeddings`] method.

        Args:
            new_tokens (`str`, `AddedToken` or a list of *str* or `AddedToken`):
                Tokens are only added if they are not already in the vocabulary. `AddedToken` wraps a string
                token to let you personalize its behavior: whether this token should only match against a single word,
                whether this token should strip all potential whitespaces on the left side, whether this token should
                strip all potential whitespaces on the right side, etc.
            special_tokens (`bool`, *optional*, defaults to `False`):
                Can be used to specify if the token is a special token. This mostly change the normalization behavior
                (special tokens like CLS or [MASK] are usually not lower-cased for instance).

        Returns:
            `int`: Number of tokens added to the vocabulary.

        Examples:

        ```python
        # Let's see how to increase the vocabulary of Bert model and tokenizer
        tokenizer = BertTokenizer.from_pretrained("bert-base-uncased")
        model = BertModel.from_pretrained("bert-base-uncased")

        num_added_toks = tokenizer.add_tokens(["new_tok1", "my_new-tok2"])
        print("We have added", num_added_toks, "tokens")
        # Notice: resize_token_embeddings expect to receive the full size of the new vocabulary, i.e., the length of the tokenizer.
        model.resize_token_embeddings(len(tokenizer))
        ```"""
        if not new_tokens:
            return 0

        if not isinstance(new_tokens, (list, tuple)):
            new_tokens = [new_tokens]

        return self._add_tokens(new_tokens, special_tokens=special_tokens)

    @classmethod
    def _add_extra_special_tokens(cls, extra_sp_token: Union[str, AddedToken]):
        if extra_sp_token not in cls.SPECIAL_TOKENS_ATTRIBUTES:
            cls.SPECIAL_TOKENS_ATTRIBUTES.append(extra_sp_token)

    def _add_tokens(
        self,
        new_tokens: Union[List[str], List[AddedToken]],
        special_tokens: bool = False,
    ) -> int:
        raise NotImplementedError

    @property
    def bos_token(self) -> str:
        """
        `str`: Beginning of sentence token. Log an error if used while not having been set.
        """
        if self._bos_token is None and self.verbose:
            logging.error("Using bos_token, but it is not set yet.")
            return None
        return str(self._bos_token)

    @property
    def eos_token(self) -> str:
        """
        `str`: End of sentence token. Log an error if used while not having been set.
        """
        if self._eos_token is None and self.verbose:
            logging.error("Using eos_token, but it is not set yet.")
            return None
        return str(self._eos_token)

    @property
    def unk_token(self) -> str:
        """
        `str`: Unknown token. Log an error if used while not having been set.
        """
        if self._unk_token is None and self.verbose:
            logging.error("Using unk_token, but it is not set yet.")
            return None
        return str(self._unk_token)

    @property
    def sep_token(self) -> str:
        """
        `str`: Separation token, to separate context and query in an input sequence. Log an error if used while not
        having been set.
        """
        if self._sep_token is None and self.verbose:
            logging.error("Using sep_token, but it is not set yet.")
            return None
        return str(self._sep_token)

    @property
    def pad_token(self) -> str:
        """
        `str`: Padding token. Log an error if used while not having been set.
        """
        if self._pad_token is None and self.verbose:
            logging.error("Using pad_token, but it is not set yet.")
            return None
        return str(self._pad_token)

    @property
    def cls_token(self) -> str:
        """
        `str`: Classification token, to extract a summary of an input sequence leveraging self-attention along the full
        depth of the model. Log an error if used while not having been set.
        """
        if self._cls_token is None and self.verbose:
            logging.error("Using cls_token, but it is not set yet.")
            return None
        return str(self._cls_token)

    @property
    def mask_token(self) -> str:
        """
        `str`: Mask token, to use when training a model with masked-language modeling. Log an error if used while not
        having been set.
        """
        if self._mask_token is None and self.verbose:
            logging.error("Using mask_token, but it is not set yet.")
            return None
        return str(self._mask_token)

    @property
    def additional_special_tokens(self) -> List[str]:
        """
        `List[str]`: All the additional special tokens you may want to use. Log an error if used while not having been
        set.
        """
        if self._additional_special_tokens is None and self.verbose:
            logging.error("Using additional_special_tokens, but it is not set yet.")
            return None
        return [str(tok) for tok in self._additional_special_tokens]

    @bos_token.setter
    def bos_token(self, value):
        self._bos_token = value

    @eos_token.setter
    def eos_token(self, value):
        self._eos_token = value

    @unk_token.setter
    def unk_token(self, value):
        self._unk_token = value

    @sep_token.setter
    def sep_token(self, value):
        self._sep_token = value

    @pad_token.setter
    def pad_token(self, value):
        self._pad_token = value

    @cls_token.setter
    def cls_token(self, value):
        self._cls_token = value

    @mask_token.setter
    def mask_token(self, value):
        self._mask_token = value

    @additional_special_tokens.setter
    def additional_special_tokens(self, value):
        self._additional_special_tokens = value

    @property
    def bos_token_id(self) -> Optional[int]:
        """
        `Optional[int]`: Id of the beginning of sentence token in the vocabulary. Returns `None` if the token has not
        been set.
        """
        if self._bos_token is None:
            return None
        return self.convert_tokens_to_ids(self.bos_token)

    @property
    def eos_token_id(self) -> Optional[int]:
        """
        `Optional[int]`: Id of the end of sentence token in the vocabulary. Returns `None` if the token has not been
        set.
        """
        if self._eos_token is None:
            return None
        return self.convert_tokens_to_ids(self.eos_token)

    @property
    def unk_token_id(self) -> Optional[int]:
        """
        `Optional[int]`: Id of the unknown token in the vocabulary. Returns `None` if the token has not been set.
        """
        if self._unk_token is None:
            return None
        return self.convert_tokens_to_ids(self.unk_token)

    @property
    def sep_token_id(self) -> Optional[int]:
        """
        `Optional[int]`: Id of the separation token in the vocabulary, to separate context and query in an input
        sequence. Returns `None` if the token has not been set.
        """
        if self._sep_token is None:
            return None
        return self.convert_tokens_to_ids(self.sep_token)

    @property
    def pad_token_id(self) -> Optional[int]:
        """
        `Optional[int]`: Id of the padding token in the vocabulary. Returns `None` if the token has not been set.
        """
        if self._pad_token is None:
            return None
        return self.convert_tokens_to_ids(self.pad_token)

    @property
    def pad_token_type_id(self) -> int:
        """
        `int`: Id of the padding token type in the vocabulary.
        """
        return self._pad_token_type_id

    @property
    def cls_token_id(self) -> Optional[int]:
        """
        `Optional[int]`: Id of the classification token in the vocabulary, to extract a summary of an input sequence
        leveraging self-attention along the full depth of the model.

        Returns `None` if the token has not been set.
        """
        if self._cls_token is None:
            return None
        return self.convert_tokens_to_ids(self.cls_token)

    @property
    def mask_token_id(self) -> Optional[int]:
        """
        `Optional[int]`: Id of the mask token in the vocabulary, used when training a model with masked-language
        modeling. Returns `None` if the token has not been set.
        """
        if self._mask_token is None:
            return None
        return self.convert_tokens_to_ids(self.mask_token)

    @property
    def additional_special_tokens_ids(self) -> List[int]:
        """
        `List[int]`: Ids of all the additional special tokens in the vocabulary. Log an error if used while not having
        been set.
        """
        return self.convert_tokens_to_ids(self.additional_special_tokens)

    @bos_token_id.setter
    def bos_token_id(self, value):
        self._bos_token = (
            self.convert_ids_to_tokens(value) if value is not None else None
        )

    @eos_token_id.setter
    def eos_token_id(self, value):
        self._eos_token = (
            self.convert_ids_to_tokens(value) if value is not None else None
        )

    @unk_token_id.setter
    def unk_token_id(self, value):
        self._unk_token = (
            self.convert_ids_to_tokens(value) if value is not None else None
        )

    @sep_token_id.setter
    def sep_token_id(self, value):
        self._sep_token = (
            self.convert_ids_to_tokens(value) if value is not None else None
        )

    @pad_token_id.setter
    def pad_token_id(self, value):
        self._pad_token = (
            self.convert_ids_to_tokens(value) if value is not None else None
        )

    @cls_token_id.setter
    def cls_token_id(self, value):
        self._cls_token = (
            self.convert_ids_to_tokens(value) if value is not None else None
        )

    @mask_token_id.setter
    def mask_token_id(self, value):
        self._mask_token = (
            self.convert_ids_to_tokens(value) if value is not None else None
        )

    @additional_special_tokens_ids.setter
    def additional_special_tokens_ids(self, values):
        self._additional_special_tokens = [
            self.convert_ids_to_tokens(value) for value in values
        ]

    @property
    def special_tokens_map(self) -> Dict[str, Union[str, List[str]]]:
        """
        `Dict[str, Union[str, List[str]]]`: A dictionary mapping special token class attributes (`cls_token`,
        `unk_token`, etc.) to their values (`'<unk>'`, `'<cls>'`, etc.).

        Convert potential tokens of `AddedToken` type to string.
        """
        set_attr = {}
        for attr in self.SPECIAL_TOKENS_ATTRIBUTES:
            try:
                attr_value = getattr(self, "_" + attr)
            except:
                try:
                    attr_value = getattr(self, attr)
                except:
                    continue
            if attr_value:
                set_attr[attr] = (
                    type(attr_value)(
                        str(attr_value_sub) for attr_value_sub in attr_value
                    )
                    if isinstance(attr_value, (list, tuple))
                    else str(attr_value)
                )
        return set_attr

    @property
    def special_tokens_map_extended(
        self,
    ) -> Dict[str, Union[str, AddedToken, List[Union[str, AddedToken]]]]:
        """
        `Dict[str, Union[str, AddedToken, List[Union[str, AddedToken]]]]`: A dictionary mapping
        special token class attributes (`cls_token`, `unk_token`, etc.) to their values (`'<unk>'`, `'<cls>'`, etc.).

        Don't convert tokens of `AddedToken` type to string so they can be used to control more finely how
        special tokens are tokenized.
        """
        set_attr = {}
        for attr in self.SPECIAL_TOKENS_ATTRIBUTES:
            try:
                attr_value = getattr(self, "_" + attr)
            except:
                try:
                    attr_value = getattr(self, attr)
                except:
                    continue
            if attr_value:
                set_attr[attr] = attr_value
        return set_attr

    @property
    def all_special_tokens(self) -> List[str]:
        """
        `List[str]`: All the special tokens (`'<unk>'`, `'<cls>'`, etc.) mapped to class attributes.

        Convert tokens of `AddedToken` type to string.
        """
        all_toks = [str(s) for s in self.all_special_tokens_extended]
        return all_toks

    @property
    def all_special_tokens_extended(self) -> List[Union[str, AddedToken]]:
        """
        `List[Union[str, AddedToken]]`: All the special tokens (`'<unk>'`, `'<cls>'`, etc.) mapped to class
        attributes.

        Don't convert tokens of `AddedToken` type to string so they can be used to control more finely how
        special tokens are tokenized.
        """
        all_tokens = []
        seen = set()
        for value in self.special_tokens_map_extended.values():
            if isinstance(value, (list, tuple)):
                tokens_to_add = [token for token in value if str(token) not in seen]
            else:
                tokens_to_add = [value] if str(value) not in seen else []
            seen.update(map(str, tokens_to_add))
            all_tokens.extend(tokens_to_add)
        return all_tokens

    @property
    def all_special_ids(self) -> List[int]:
        """
        `List[int]`: List the ids of the special tokens(`'<unk>'`, `'<cls>'`, etc.) mapped to class attributes.
        """
        all_toks = self.all_special_tokens
        all_ids = self.convert_tokens_to_ids(all_toks)
        return all_ids


class PretrainedTokenizerBase(SpecialTokensMixin):
    """
    Base class for [`PretrainedTokenizer`].

    Class attributes (overridden by derived classes)

         - **resource_files_names** (`Dict[str, str]`) -- A dictionary with, as keys, the `__init__` keyword name of each
            vocabulary file required by the model, and as associated values, the filename for saving the associated file
            (string).
        - **pretrained_resource_files_map** (`Dict[str, Dict[str, str]]`) -- A dictionary of dictionaries, with the
            high-level keys being the `__init__` keyword name of each vocabulary file required by the model, the
            low-level being the `short-cut-names` of the pretrained models with, as associated values, the `url` to the
            associated pretrained vocabulary file.
        - **max_model_input_sizes** (`Dict[str, Optional[int]]`) -- A dictionary with, as keys, the `short-cut-names`
            of the pretrained models, and as associated values, the maximum length of the sequence inputs of this model,
            or `None` if the model has no maximum input size.
        - **pretrained_init_configuration** (`Dict[str, Dict[str, Any]]`) -- A dictionary with, as keys, the
            `short-cut-names` of the pretrained models, and as associated values, a dictionary of specific arguments to
            pass to the `__init__` method of the tokenizer class for this pretrained model when loading the tokenizer
            with the [`~tokenizer_utils_base.PretrainedTokenizerBase.from_pretrained`] method.
        - **model_input_names** (`List[str]`) -- A list of inputs expected in the forward pass of the model.
        - **padding_side** (`str`) -- The default value for the side on which the model should have padding applied.
            Should be `'right'` or `'left'`.
        - **truncation_side** (`str`) -- The default value for the side on which the model should have truncation
            applied. Should be `'right'` or `'left'`.

    Args:
        model_max_length (`int`, *optional*):
            The maximum length (in number of tokens) for the inputs to the transformer model. When the tokenizer is
            loaded with [`~tokenizer_utils_base.PretrainedTokenizerBase.from_pretrained`], this will be set to the
            value stored for the associated model in `max_model_input_sizes` (see above). If no value is provided, will
            default to VERY_LARGE_INTEGER (`int(1e30)`).
        padding_side (`str`, *optional*):
            The side on which the model should have padding applied. Should be selected between ['right', 'left'].
            Default value is picked from the class attribute of the same name.
        truncation_side (`str`, *optional*):
            The side on which the model should have truncation applied. Should be selected between ['right', 'left'].
            Default value is picked from the class attribute of the same name.
        model_input_names (`List[string]`, *optional*):
            The list of inputs accepted by the forward pass of the model (like `"token_type_ids"` or
            `"attention_mask"`). Default value is picked from the class attribute of the same name.
        bos_token (`str` or `AddedToken`, *optional*):
            A special token representing the beginning of a sentence. Will be associated to `self.bos_token` and
            `self.bos_token_id`.
        eos_token (`str` or `AddedToken`, *optional*):
            A special token representing the end of a sentence. Will be associated to `self.eos_token` and
            `self.eos_token_id`.
        unk_token (`str` or `AddedToken`, *optional*):
            A special token representing an out-of-vocabulary token. Will be associated to `self.unk_token` and
            `self.unk_token_id`.
        sep_token (`str` or `AddedToken`, *optional*):
            A special token separating two different sentences in the same input (used by BERT for instance). Will be
            associated to `self.sep_token` and `self.sep_token_id`.
        pad_token (`str` or `AddedToken`, *optional*):
            A special token used to make arrays of tokens the same size for batching purpose. Will then be ignored by
            attention mechanisms or loss computation. Will be associated to `self.pad_token` and `self.pad_token_id`.
        cls_token (`str` or `AddedToken`, *optional*):
            A special token representing the class of the input (used by BERT for instance). Will be associated to
            `self.cls_token` and `self.cls_token_id`.
        mask_token (`str` or `AddedToken`, *optional*):
            A special token representing a masked token (used by masked-language modeling pretraining objectives, like
            BERT). Will be associated to `self.mask_token` and `self.mask_token_id`.
        additional_special_tokens (tuple or list of `str` or `AddedToken`, *optional*):
            A tuple or a list of additional special tokens. Add them here to ensure they won't be split by the
            tokenization process. Will be associated to `self.additional_special_tokens` and
            `self.additional_special_tokens_ids`.
    """

    resource_files_names: Dict[str, str] = {}
    pretrained_resource_files_map: Dict[str, Dict[str, str]] = {}
    pretrained_init_configuration: Dict[str, Dict[str, Any]] = {}
    max_model_input_sizes: Dict[str, Optional[int]] = {}
    _auto_class: Optional[str] = None
    tokenizer_config_file = TOKENIZER_CONFIG_NAME

    # first name has to correspond to main model input name
    # to make sure `tokenizer.pad(...)` works correctly
    model_input_names: List[str] = ["input_ids", "token_type_ids"]
    padding_side: str = "right"
    truncation_side: str = "right"
    slow_tokenizer_class = None

    def __init__(self, **kwargs):
        # inputs and kwargs for saving and re-loading (see ``from_pretrained`` and ``save_pretrained``)
        self.init_inputs = ()

        self.init_kwargs = getattr(self, "init_kwargs", None) or copy.deepcopy(kwargs)
        self.name_or_path = kwargs.pop("name_or_path", "")
        self._processor_class = kwargs.pop("processor_class", None)

        # For backward compatibility we fallback to set model_max_length from max_len if provided
        model_max_length = kwargs.pop("model_max_length", kwargs.pop("max_len", None))
        self.model_max_length = (
            model_max_length if model_max_length is not None else VERY_LARGE_INTEGER
        )

        # Padding and truncation side are right by default and overridden in subclasses. If specified in the kwargs, it
        # is changed.
        self.padding_side = kwargs.pop("padding_side", self.padding_side)
        if self.padding_side not in ["right", "left"]:
            raise ValueError(
                f"Padding side should be selected between 'right' and 'left', current value: {self.padding_side}"
            )

        self.truncation_side = kwargs.pop("truncation_side", self.truncation_side)
        if self.truncation_side not in ["right", "left"]:
            raise ValueError(
                f"Padding side should be selected between 'right' and 'left', current value: {self.truncation_side}"
            )

        self.model_input_names = kwargs.pop("model_input_names", self.model_input_names)

        self.clean_up_tokenization_spaces = kwargs.pop(
            "clean_up_tokenization_spaces", False
        )

        self.split_special_tokens = kwargs.pop("split_special_tokens", False)

        self.deprecation_warnings = (
            {}
        )  # Use to store when we have already noticed a deprecation warning (avoid overlogging).

        super().__init__(**kwargs)

    @property
    def max_len_single_sentence(self) -> int:
        """
        `int`: The maximum length of a sentence that can be fed to the model.
        """
        return self.model_max_length - self.num_special_tokens_to_add(pair=False)

    @property
    def max_len_sentences_pair(self) -> int:
        """
        `int`: The maximum combined length of a pair of sentences that can be fed to the model.
        """
        return self.model_max_length - self.num_special_tokens_to_add(pair=True)

    @max_len_single_sentence.setter
    def max_len_single_sentence(self, value) -> int:
        # For backward compatibility, allow to try to setup 'max_len_single_sentence'.
        if (
            value == self.model_max_length - self.num_special_tokens_to_add(pair=False)
            and self.verbose
        ):
            if not self.deprecation_warnings.get("max_len_single_sentence", False):
                warnings.warn(
                    "Setting 'max_len_single_sentence' is now deprecated. "
                    "This value is automatically set up."
                )
            self.deprecation_warnings["max_len_single_sentence"] = True
        else:
            raise ValueError(
                "Setting 'max_len_single_sentence' is now deprecated. "
                "This value is automatically set up."
            )

    def _switch_to_input_mode(self):
        """
        Private method to put the tokenizer in input mode (when it has different modes for input/outputs)
        """
        pass

    @max_len_sentences_pair.setter
    def max_len_sentences_pair(self, value) -> int:
        if (
            value == self.model_max_length - self.num_special_tokens_to_add(pair=True)
            and self.verbose
        ):
            if not self.deprecation_warnings.get("max_len_sentences_pair", False):
                warnings.warn(
                    "Setting 'max_len_sentences_pair' is now deprecated. "
                    "This value is automatically set up."
                )
            self.deprecation_warnings["max_len_sentences_pair"] = True
        else:
            raise ValueError(
                "Setting 'max_len_sentences_pair' is now deprecated. "
                "This value is automatically set up."
            )

    def _set_processor_class(self, processor_class: str):
        """Sets processor class as an attribute."""
        self._processor_class = processor_class

    def __repr__(self) -> str:
        added_tokens_decoder_rep = "\n\t".join(
            [f"{k}: {v.__repr__()}," for k, v in self.added_tokens_decoder.items()]
        )
        return (
            f"{self.__class__.__name__}(name_or_path='{self.name_or_path}',"
            f" vocab_size={self.vocab_size}, model_max_length={self.model_max_length}, is_fast={self.is_fast},"
            f" padding_side='{self.padding_side}', truncation_side='{self.truncation_side}',"
            f" special_tokens={self.special_tokens_map}, clean_up_tokenization_spaces={self.clean_up_tokenization_spaces}), "
            " added_tokens_decoder={\n\t" + added_tokens_decoder_rep + "\n}"
        )

    def get_vocab(self) -> Dict[str, int]:
        """
        Returns the vocabulary as a dictionary of token to index.

        `tokenizer.get_vocab()[token]` is equivalent to `tokenizer.convert_tokens_to_ids(token)` when `token` is in the
        vocab.

        Returns:
            `Dict[str, int]`: The vocabulary.
        """
        raise NotImplementedError()

    @classmethod
    def from_pretrained(cls, pretrained_model_name_or_path, *args, **kwargs):
        """
        Creates an instance of `PretrainedTokenizer`. Related resources are loaded
        by specifying name of a built-in pretrained model, or a community-contributed
        pretrained model, or a local file directory path.

        Args:
            pretrained_model_name_or_path (str): Name of pretrained model or dir path
                to load from. The string can be:

                - Name of built-in pretrained model
                - Name of a community-contributed pretrained model.
                - Local directory path which contains tokenizer related resources
                  and tokenizer config file ("tokenizer_config.json").
            from_hf_hub (bool, optional): whether to load from Huggingface Hub
            subfolder (str, optional) An optional value corresponding to a folder inside the repo.
                Only works when loading from Huggingface Hub.
            *args (tuple): position arguments for model `__init__`. If provided,
                use these as position argument values for tokenizer initialization.
            **kwargs (dict): keyword arguments for model `__init__`. If provided,
                use these to update pre-defined keyword argument values for tokenizer
                initialization.

        Returns:
            PretrainedTokenizer: An instance of `PretrainedTokenizer`.

        Example:
            .. code-block::

                from paddlenlp.transformers import BertTokenizer

                # Name of built-in pretrained model
                tokenizer = BertTokenizer.from_pretrained('bert-base-uncased')

                # Name of community-contributed pretrained model
                tokenizer = BertTokenizer.from_pretrained('yingyibiao/bert-base-uncased-sst-2-finetuned')

                # Load from local directory path
                tokenizer = BertTokenizer.from_pretrained('./my_bert/')
        """
        cache_dir = kwargs.pop("cache_dir", None)
        from_hf_hub = kwargs.pop("from_hf_hub", False)
        from_aistudio = kwargs.pop("from_aistudio", False)
        subfolder = kwargs.pop("subfolder", "")
        return_tokenizer_file_dir = kwargs.pop("return_tokenizer_file_dir", False)

        pretrained_model_name_or_path = str(pretrained_model_name_or_path)
        vocab_files = {}
        init_configuration = {}

        additional_files_names = {
            "added_tokens_file": ADDED_TOKENS_FILE,
            "special_tokens_map_file": SPECIAL_TOKENS_MAP_FILE,
            "tokenizer_config_file": TOKENIZER_CONFIG_FILE,
            "chat_template_file": CHAT_TEMPLATE_CONFIG_NAME,
        }

        if hasattr(cls, "vocab_files_names") and len(cls.resource_files_names) == 0:
            cls.resource_files_names = copy.deepcopy(cls.vocab_files_names)
            logging.error(
                "The attribute 'vocab_files_names' is deprecated. Please use 'resource_files_names' instead.",
                DeprecationWarning,
            )
        vocab_files_target = {**cls.resource_files_names, **additional_files_names}
        # From HF Hub or AI Studio
        if from_hf_hub or from_aistudio:
            # Only include the necessary resource files specified by the tokenizer cls
            # Deep copy to avoid modifying the class attributes
            vocab_files = copy.deepcopy(cls.resource_files_names)
            vocab_files["tokenizer_config_file"] = cls.tokenizer_config_file

        # From built-in pretrained models
        elif pretrained_model_name_or_path in cls.pretrained_init_configuration:
            for file_id, map_list in cls.pretrained_resource_files_map.items():
                vocab_files[file_id] = map_list[pretrained_model_name_or_path]
            init_configuration = copy.deepcopy(
                cls.pretrained_init_configuration[pretrained_model_name_or_path]
            )
        # From local dir path
        elif os.path.isdir(pretrained_model_name_or_path):
            vocab_files_target["tokenizer_config_file"] = cls.tokenizer_config_file
            for file_id, file_name in vocab_files_target.items():
                full_file_name = os.path.join(
                    pretrained_model_name_or_path, subfolder, file_name
                )
                if os.path.isfile(full_file_name):
                    vocab_files[file_id] = full_file_name
        else:
            # Assuming from community-contributed pretrained models
            for file_id, file_name in vocab_files_target.items():
                vocab_files[file_id] = file_name
        resolved_vocab_files = {}
        for file_id, file_path in vocab_files.items():
            # adapt to PaddleX
            resolved_vocab_files[file_id] = file_path

        for file_id, file_path in resolved_vocab_files.items():
            if resolved_vocab_files[file_id] is not None:
                cache_dir = os.path.dirname(resolved_vocab_files[file_id])
                break
        return cls._from_pretrained(
            resolved_vocab_files,
            pretrained_model_name_or_path,
            init_configuration,
            *args,
            cache_dir=cache_dir,
            return_tokenizer_file_dir=return_tokenizer_file_dir,
            from_hf_hub=from_hf_hub,
            **kwargs,
        )

    @classmethod
    def _from_pretrained(
        cls,
        resolved_vocab_files,
        pretrained_model_name_or_path,
        init_configuration,
        *init_inputs,
        cache_dir=None,
        return_tokenizer_file_dir=False,
        from_hf_hub=False,
        **kwargs,
    ):
        if cls.__name__.endswith("Fast"):
            from_slow = kwargs.get("from_slow", False)
        else:
            from_slow = kwargs.get("from_slow", True)
        has_tokenizer_file = (
            resolved_vocab_files.get("tokenizer_file", None) is not None
        )
        if (
            from_slow or not has_tokenizer_file
        ) and cls.slow_tokenizer_class is not None:
            slow_tokenizer = (cls.slow_tokenizer_class)._from_pretrained(
                copy.deepcopy(resolved_vocab_files),
                pretrained_model_name_or_path,
                copy.deepcopy(init_configuration),
                *init_inputs,
                cache_dir=cache_dir,
                **(copy.deepcopy(kwargs)),
            )
        else:
            slow_tokenizer = None
        tokenizer_config_file_dir_list = set()
        for k, v in resolved_vocab_files.items():
            if v is not None and os.path.isfile(v):
                tokenizer_config_file_dir_list.add(os.path.dirname(v))
        tokenizer_config_file_dir_list = list(tokenizer_config_file_dir_list)
        # TODO: check this
        assert (
            len(tokenizer_config_file_dir_list) > 0
        ), "All tokenizer files should be in the same directory."

        has_tokenizer_file = (
            resolved_vocab_files.get("tokenizer_file", None) is not None
        )
        tokenizer_config_file = resolved_vocab_files.pop("tokenizer_config_file", None)
        if tokenizer_config_file is not None:
            with io.open(tokenizer_config_file, encoding="utf-8") as f:
                init_kwargs = json.load(f)
            init_kwargs.pop("tokenizer_class", None)
        else:
            init_kwargs = init_configuration

        if slow_tokenizer is not None:
            init_kwargs["__slow_tokenizer"] = slow_tokenizer
        init_kwargs["name_or_path"] = pretrained_model_name_or_path
        init_kwargs["from_slow"] = from_slow

        pass_added_tokens_file = False
        added_tokens_decoder: Dict[int, AddedToken] = {}
        if "added_tokens_decoder" in init_kwargs:
            for idx, token in init_kwargs["added_tokens_decoder"].items():
                if isinstance(token, dict):
                    token = AddedToken(**token)
                if isinstance(token, AddedToken):
                    added_tokens_decoder[int(idx)] = token
                else:
                    raise ValueError(
                        f"Found a {token.__class__} in the saved `added_tokens_decoder`, should be a dictionary or an AddedToken instance"
                    )
            init_kwargs["added_tokens_decoder"] = (
                added_tokens_decoder  # NOTE tokenizer_config.json下, 注册的`added_tokens_decoder`被解析成字典
            )
            pass_added_tokens_file = True

        init_kwargs.pop("init_class", None)

        init_kwargs.update(kwargs)

        def convert_added_tokens(obj):
            if (
                isinstance(obj, dict)
                and "__type" in obj
                and obj["__type"] == "AddedToken"
            ):
                obj.pop("__type")
                return AddedToken(**obj)
            elif isinstance(obj, (list, tuple)):
                return list(convert_added_tokens(o) for o in obj)
            elif isinstance(obj, dict):
                return {k: convert_added_tokens(v) for k, v in obj.items()}
            return obj

        init_kwargs = convert_added_tokens(init_kwargs)
        if pretrained_model_name_or_path in cls.max_model_input_sizes:
            model_max_length = cls.max_model_input_sizes[pretrained_model_name_or_path]
            if model_max_length is not None and isinstance(
                model_max_length, (int, float)
            ):
                init_kwargs["model_max_length"] = min(
                    init_kwargs.get("model_max_length", int(1e30)), model_max_length
                )

        for args_name, file_path in resolved_vocab_files.items():
            if args_name not in init_kwargs or init_kwargs[args_name] is None:
                init_kwargs[args_name] = file_path
            elif not os.path.isfile(init_kwargs[args_name] or "") and os.path.isfile(
                file_path
            ):
                init_kwargs[args_name] = file_path

        if from_hf_hub and "tokenizer_file" in init_kwargs:
            init_kwargs.pop("tokenizer_file")

        try:
            tokenizer = cls(*init_inputs, **init_kwargs)
        # adapt to PaddleX
        except RuntimeError as e:
            if "sentencepiece_processor.cc" in str(e):
                logging.info(
                    "Unable to load tokenizer model from SPM, loading from TikToken will be attempted instead."
                    "(SentencePiece RuntimeError: Tried to load SPM model with non-SPM vocab file).",
                )
            return False

        chat_template = init_kwargs.pop("chat_template", None)
        if chat_template is not None:
            tokenizer.init_chat_template(chat_template)
        special_tokens_map_file = resolved_vocab_files.pop(
            "special_tokens_map_file", None
        )
        if special_tokens_map_file is not None:
            with open(
                special_tokens_map_file, encoding="utf-8"
            ) as special_tokens_map_handle:
                special_tokens_map = json.load(special_tokens_map_handle)
            for key, value in special_tokens_map.items():
                if key in kwargs and kwargs[key]:
                    continue
                if isinstance(value, dict):
                    value = AddedToken(**value)
                elif isinstance(value, list):
                    value = [
                        AddedToken(**token) if isinstance(token, dict) else token
                        for token in value
                    ]
                setattr(tokenizer, key, value)
                cls._add_extra_special_tokens(key)

        special_tokens = tokenizer.all_special_tokens
        added_tokens_file = resolved_vocab_files.pop("added_tokens_file", None)
        added_tokens_file = None if pass_added_tokens_file else added_tokens_file
        if added_tokens_file is not None:
            with open(added_tokens_file, encoding="utf-8") as added_tokens_handle:
                added_tok_encoder = json.load(added_tokens_handle)

            added_tok_encoder_sorted = list(
                sorted(added_tok_encoder.items(), key=lambda x: x[1])
            )
            for token, index in added_tok_encoder_sorted:
                if (
                    has_tokenizer_file
                    and index != len(tokenizer)
                    and tokenizer.convert_tokens_to_ids(token) != index
                ):
                    raise ValueError(
                        f"Wrong index found for {token}: should be {tokenizer.convert_tokens_to_ids(token)} but found "
                        f"{index}."
                    )
                elif not has_tokenizer_file and index != len(tokenizer):
                    raise ValueError(
                        f"Non-consecutive added token '{token}' found. "
                        f"Should have index {len(tokenizer)} but has index {index} in saved vocabulary."
                    )

                tokenizer.add_tokens(
                    token, special_tokens=bool(token in special_tokens)
                )
        added_tokens = tokenizer.sanitize_special_tokens()
        if added_tokens:
            logging.info(
                "Special tokens have been added in the vocabulary, make sure the associated word embeddings are fine-tuned or trained."
            )
        if pretrained_model_name_or_path in cls.pretrained_init_configuration:
            tokenizer.save_pretrained(cache_dir)

        if return_tokenizer_file_dir:
            return tokenizer, list(tokenizer_config_file_dir_list)[0]
        return tokenizer

    def save_pretrained(
        self, save_directory, filename_prefix: Optional[str] = None, **kwargs
    ):
        """
        Save tokenizer configuration and related resources to files under
        `save_directory`. The tokenizer configuration would be saved into
        `tokenizer_config_file` indicating file (thus `tokenizer_config.json`),
        and resources would be saved into `resource_files_names` indicating files
        by using `self.save_resources(save_directory)`.

        The `save_directory` can be used in `from_pretrained` as argument value
        of `pretrained_model_name_or_path` to re-load the tokenizer.

        Args:
            save_directory (str): Directory to save files into.
            filename_prefix: (str, optional):
                A prefix to add to the names of the files saved by the tokenizer.

        Example:
            .. code-block::

                from paddlenlp.transformers import BertTokenizer

                tokenizer = BertTokenizer.from_pretrained('bert-base-uncased')
                tokenizer.save_pretrained('trained_model')
                # reload from save_directory
                tokenizer = BertTokenizer.from_pretrained('trained_model')
        """
        assert not os.path.isfile(
            save_directory
        ), "Saving directory ({}) should be a directory, not a file".format(
            save_directory
        )
        os.makedirs(save_directory, exist_ok=True)

        special_tokens_map_file = os.path.join(
            save_directory,
            (filename_prefix + "-" if filename_prefix else "")
            + SPECIAL_TOKENS_MAP_FILE,
        )
        tokenizer_config_file = os.path.join(
            save_directory,
            (filename_prefix + "-" if filename_prefix else "")
            + self.tokenizer_config_file,
        )

        tokenizer_config = copy.deepcopy(self.init_kwargs)
        if len(self.init_inputs) > 0:
            tokenizer_config["init_inputs"] = copy.deepcopy(self.init_inputs)
        for file_id in self.resource_files_names.keys():
            tokenizer_config.pop(file_id, None)

        def convert_added_tokens(obj: Union[AddedToken, Any], add_type_field=True):
            if isinstance(obj, AddedToken):
                out = obj.__getstate__()
                if add_type_field:
                    out["__type"] = "AddedToken"
                return out
            elif isinstance(obj, (list, tuple)):
                return list(
                    convert_added_tokens(o, add_type_field=add_type_field) for o in obj
                )
            elif isinstance(obj, dict):
                return {
                    k: convert_added_tokens(v, add_type_field=add_type_field)
                    for k, v in obj.items()
                }
            return obj

        tokenizer_config = convert_added_tokens(tokenizer_config, add_type_field=True)

        added_tokens = {}
        for key, value in self.added_tokens_decoder.items():
            if isinstance(value, AddedToken):
                added_tokens[key] = value.__getstate__()
            else:
                added_tokens[key] = AddedToken(value).__getstate__()
        tokenizer_config["added_tokens_decoder"] = added_tokens

        tokenizer_class = self.__class__.__name__
        tokenizer_config["tokenizer_class"] = tokenizer_class

        with io.open(tokenizer_config_file, "w", encoding="utf-8") as f:
            f.write(json.dumps(tokenizer_config, ensure_ascii=False))
        logging.info(f"tokenizer config file saved in {tokenizer_config_file}")

        write_dict = convert_added_tokens(
            self.special_tokens_map_extended, add_type_field=False
        )
        with open(special_tokens_map_file, "w", encoding="utf-8") as f:
            f.write(json.dumps(write_dict, ensure_ascii=False))
        logging.info(f"Special tokens file saved in {special_tokens_map_file}")

        file_names = (tokenizer_config_file, special_tokens_map_file)

        save_files = self._save_pretrained(
            save_directory=save_directory,
            file_names=file_names,
            filename_prefix=filename_prefix,
        )

        return save_files

    def _save_pretrained(
        self,
        save_directory: Union[str, os.PathLike],
        file_names: Tuple[str],
        filename_prefix: Optional[str] = None,
    ) -> Tuple[str]:
        """
        Save a tokenizer using the tokenizer format: vocabulary + added tokens.

        """
        save_directory = str(save_directory)

        added_tokens_file = os.path.join(
            save_directory,
            (filename_prefix + "-" if filename_prefix else "") + ADDED_TOKENS_FILE,
        )
        added_vocab = self.get_added_vocab()
        if added_vocab:
            with open(added_tokens_file, "w", encoding="utf-8") as f:
                out_str = json.dumps(added_vocab, ensure_ascii=False)
                f.write(out_str)
                logging.info(f"added tokens file saved in {added_tokens_file}")

        self.save_resources(save_directory)

        return file_names + (added_tokens_file,)

    def tokenize(
        self,
        text: str,
        pair: Optional[str] = None,
        add_special_tokens: bool = False,
        **kwargs,
    ) -> List[str]:
        """
        Converts a string in a sequence of tokens, replacing unknown tokens with the `unk_token`.

        Args:
            text (`str`):
                The sequence to be encoded.
            pair (`str`, *optional*):
                A second sequence to be encoded with the first.
            add_special_tokens (`bool`, *optional*, defaults to `False`):
                Whether or not to add the special tokens associated with the corresponding model.
            kwargs (additional keyword arguments, *optional*):
                Will be passed to the underlying model specific encode method. See details in
                [`~PretrainedTokenizerBase.__call__`]

        Returns:
            `List[str]`: The list of tokens.
        """
        raise NotImplementedError

    def num_special_tokens_to_add(self, pair: bool = False) -> int:
        raise NotImplementedError

    def _get_padding_truncation_strategies(
        self,
        padding=False,
        truncation=False,
        max_length=None,
        pad_to_multiple_of=None,
        verbose=True,
        **kwargs,
    ):
        """
        Find the correct padding/truncation strategy with backward compatibility for old arguments (truncation_strategy
        and pad_to_max_length) and behaviors.
        """
        old_truncation_strategy = kwargs.pop("truncation_strategy", "do_not_truncate")
        old_pad_to_max_length = kwargs.pop("pad_to_max_seq_len", False)

        if max_length is not None and padding is False and truncation is False:
            if verbose:
                if not self.deprecation_warnings.get(
                    "Truncation-not-explicitly-activated", False
                ):
                    warnings.warn(
                        "Truncation was not explicitly activated but `max_length` is provided a specific value, "
                        "please use `truncation=True` to explicitly truncate examples to max length. "
                        "Defaulting to 'longest_first' truncation strategy. "
                        "If you encode pairs of sequences (GLUE-style) with the tokenizer you can select this strategy "
                        "more precisely by providing a specific strategy to `truncation`."
                    )
                self.deprecation_warnings["Truncation-not-explicitly-activated"] = True
            truncation = "longest_first"

        # Get padding strategy
        if padding is False and old_pad_to_max_length:
            if verbose:
                warnings.warn(
                    "The `pad_to_max_length` argument is deprecated and will be removed in a future version, "
                    "use `padding=True` or `padding='longest'` to pad to the longest sequence in the batch, or "
                    "use `padding='max_length'` to pad to a max length. In this case, you can give a specific "
                    "length with `max_length` (e.g. `max_length=45`) or leave max_length to None to pad to the "
                    "maximal input size of the model (e.g. 512 for Bert).",
                    FutureWarning,
                )
            if max_length is None:
                padding_strategy = PaddingStrategy.LONGEST
            else:
                padding_strategy = PaddingStrategy.MAX_LENGTH
        elif padding is not False:
            if padding is True:
                if verbose:
                    if max_length is not None and (
                        truncation is False or truncation == "do_not_truncate"
                    ):
                        warnings.warn(
                            "`max_length` is ignored when `padding`=`True` and there is no truncation strategy. "
                            "To pad to max length, use `padding='max_length'`."
                        )
                    if old_pad_to_max_length is not False:
                        warnings.warn(
                            "Though `pad_to_max_length` = `True`, it is ignored because `padding`=`True`."
                        )
                padding_strategy = PaddingStrategy.LONGEST
            elif not isinstance(padding, PaddingStrategy):
                padding_strategy = PaddingStrategy(padding)
            elif isinstance(padding, PaddingStrategy):
                padding_strategy = padding
        else:
            padding_strategy = PaddingStrategy.DO_NOT_PAD

        # Get truncation strategy
        if truncation is False and old_truncation_strategy != "do_not_truncate":
            if verbose:
                warnings.warn(
                    "The `truncation_strategy` argument is deprecated and will be removed in a future version, "
                    "use `truncation=True` to truncate examples to a max length. You can give a specific "
                    "length with `max_length` (e.g. `max_length=45`) or leave max_length to None to truncate to the "
                    "maximal input size of the model (e.g. 512 for Bert). "
                    " If you have pairs of inputs, you can give a specific truncation strategy selected among "
                    "`truncation='only_first'` (will only truncate the first sentence in the pairs) "
                    "`truncation='only_second'` (will only truncate the second sentence in the pairs) "
                    "or `truncation='longest_first'` (will iteratively remove tokens from the longest sentence in the pairs).",
                    FutureWarning,
                )
            truncation_strategy = TruncationStrategy(old_truncation_strategy)
        elif truncation is not False and truncation is not None:
            if truncation is True:
                truncation_strategy = (
                    TruncationStrategy.LONGEST_FIRST
                )  # Default to truncate the longest sequences in pairs of inputs
            elif not isinstance(truncation, TruncationStrategy):
                truncation_strategy = TruncationStrategy(truncation)
            elif isinstance(truncation, TruncationStrategy):
                truncation_strategy = truncation
        else:
            truncation_strategy = TruncationStrategy.DO_NOT_TRUNCATE

        # Set max length if needed
        if max_length is None:
            if padding_strategy == PaddingStrategy.MAX_LENGTH:
                if self.model_max_length > LARGE_INTEGER:
                    if verbose:
                        if not self.deprecation_warnings.get(
                            "Asking-to-pad-to-max_length", False
                        ):
                            warnings.warn(
                                "Asking to pad to max_length but no maximum length is provided and the model has no predefined maximum length. "
                                "Default to no padding."
                            )
                        self.deprecation_warnings["Asking-to-pad-to-max_length"] = True
                    padding_strategy = PaddingStrategy.DO_NOT_PAD
                else:
                    max_length = self.model_max_length

            if truncation_strategy != TruncationStrategy.DO_NOT_TRUNCATE:
                if self.model_max_length > LARGE_INTEGER:
                    if verbose:
                        if not self.deprecation_warnings.get(
                            "Asking-to-truncate-to-max_length", False
                        ):
                            warnings.warn(
                                "Asking to truncate to max_length but no maximum length is provided and the model has no predefined maximum length. "
                                "Default to no truncation."
                            )
                        self.deprecation_warnings[
                            "Asking-to-truncate-to-max_length"
                        ] = True
                    truncation_strategy = TruncationStrategy.DO_NOT_TRUNCATE
                else:
                    max_length = self.model_max_length

        # Test if we have a padding token
        if padding_strategy != PaddingStrategy.DO_NOT_PAD and (
            not self.pad_token or self.pad_token_id < 0
        ):
            raise ValueError(
                "Asking to pad but the tokenizer does not have a padding token. "
                "Please select a token to use as `pad_token` `(tokenizer.pad_token = tokenizer.eos_token e.g.)` "
                "or add a new pad token via `tokenizer.add_special_tokens({'pad_token': '[PAD]'})`."
            )

        # Check that we will truncate to a multiple of pad_to_multiple_of if both are provided
        if (
            truncation_strategy != TruncationStrategy.DO_NOT_TRUNCATE
            and padding_strategy != PaddingStrategy.DO_NOT_PAD
            and pad_to_multiple_of is not None
            and max_length is not None
            and (max_length % pad_to_multiple_of != 0)
        ):
            raise ValueError(
                f"Truncation and padding are both activated but "
                f"truncation length ({max_length}) is not a multiple of pad_to_multiple_of ({pad_to_multiple_of})."
            )

        return padding_strategy, truncation_strategy, max_length, kwargs

    def __call__(
        self,
        text: Union[str, List[str], List[List[str]]],
        text_pair: Optional[Union[str, List[str], List[List[str]]]] = None,
        max_length: Optional[int] = None,
        stride: int = 0,
        is_split_into_words: Union[bool, str] = False,
        padding: Union[bool, str, PaddingStrategy] = False,
        truncation: Union[bool, str, TruncationStrategy] = False,
        return_position_ids: bool = None,
        return_token_type_ids: Optional[bool] = None,
        return_attention_mask: Optional[bool] = None,
        return_length: bool = False,
        return_overflowing_tokens: bool = False,
        return_special_tokens_mask: bool = False,
        return_dict: bool = True,
        return_offsets_mapping: bool = False,
        add_special_tokens: bool = True,
        pad_to_multiple_of: Optional[int] = None,
        padding_side: Optional[Literal["right", "left"]] = None,
        return_tensors: Optional[Union[str, TensorType]] = None,
        verbose: bool = True,
        **kwargs,
    ):
        """
        Performs tokenization and uses the tokenized tokens to prepare model
        inputs. It supports sequence or sequence pair as input, and batch input
        is allowed. `self.encode()` or `self.batch_encode()` would be called
        separately for single or batch input depending on input format and
        `is_split_into_words` argument.

        Args:
            text (str, List[str] or List[List[str]]):
                The sequence or batch of sequences to be processed. One sequence
                is a string or a list of strings depending on whether it has been
                pretokenized. If each sequence is provided as a list of strings
                (pretokenized), you must set `is_split_into_words` as `True` to
                disambiguate with a batch of sequences.
            text_pair (str, List[str] or List[List[str]], optional):
                Same as `text` argument, while it represents for the latter
                sequence of the sequence pair.
            max_length (int, optional):
                If set to a number, will limit the total sequence returned so
                that it has a maximum length. If there are overflowing tokens,
                those overflowing tokens will be added to the returned dictionary
                when `return_overflowing_tokens` is `True`. Defaults to `None`.
            stride (int, optional):
                Only available for batch input of sequence pair and mainly for
                question answering usage. When for QA, `text` represents questions
                and `text_pair` represents contexts. If `stride` is set to a
                positive number, the context will be split into multiple spans
                where `stride` defines the number of (tokenized) tokens to skip
                from the start of one span to get the next span, thus will produce
                a bigger batch than inputs to include all spans. Moreover, 'overflow_to_sample'
                and 'offset_mapping' preserving the original example and position
                information will be added to the returned dictionary. Defaults to 0.
            is_split_into_words (Union[bool, str], optional):
                when the text is words or tokens, `is_split_into_words` should be True or `token`.
                `True`: means that the text should be words which should be tokenized.
                `token`: means that the text should be tokens which already be tokenized, so it should not be tokenized again.
            padding (bool, str or [PaddingStrategy], optional):
                Activates and controls padding. Accepts the following values:

                - `True` or `'longest'`: Pad to the longest sequence in the batch (or no padding if only a single
                  sequence if provided).
                - `'max_length'`: Pad to a maximum length specified with the argument `max_length` or to the maximum
                  acceptable input length for the model if that argument is not provided.
                - `False` or `'do_not_pad'` (default): No padding (i.e., can output a batch with sequences of different
                  lengths).
                Defaults to `False`.
            truncation (bool, str or [TruncationStrategy], optional):
                Activates and controls truncation. Accepts the following values:

                - `True` or `'longest_first'`: Truncate to a maximum length specified with the argument `max_length` or
                  to the maximum acceptable input length for the model if that argument is not provided. This will
                  truncate token by token, removing a token from the longest sequence in the pair if a pair of
                  sequences (or a batch of pairs) is provided.
                - `'only_first'`: Truncate to a maximum length specified with the argument `max_length` or to the
                  maximum acceptable input length for the model if that argument is not provided. This will only
                  truncate the first sequence of a pair if a pair of sequences (or a batch of pairs) is provided.
                - `'only_second'`: Truncate to a maximum length specified with the argument `max_length` or to the
                  maximum acceptable input length for the model if that argument is not provided. This will only
                  truncate the second sequence of a pair if a pair of sequences (or a batch of pairs) is provided.
                - `False` or `'do_not_truncate'` (default): No truncation (i.e., can output batch with sequence lengths
                  greater than the model maximum admissible input size).
                Defaults to `False`.
            return_position_ids (bool, optional):
                Whether to include tokens position ids in the returned dictionary.
                Defaults to `False`.
            return_token_type_ids (bool, optional):
                Whether to include token type ids in the returned dictionary.
                Defaults to `True`.
            return_attention_mask (bool, optional):
                Whether to include the attention mask in the returned dictionary.
                Defaults to `False`.
            return_length (bool, optional):
                Whether to include the length of each encoded inputs in the
                returned dictionary. Defaults to `False`.
            return_overflowing_tokens (bool, optional):
                Whether to include overflowing token information in the returned
                dictionary. Defaults to `False`.
            return_special_tokens_mask (bool, optional):
                Whether to include special tokens mask information in the returned
                dictionary. Defaults to `False`.
            return_dict (bool, optional):
                Decide the format for returned encoded batch inputs. Only works when
                input is a batch of data.
                ::
                    - If True, encoded inputs would be a dictionary like:
                        {'input_ids': [[1, 4444, 4385, 1545, 6712],[1, 4444, 4385]],
                        'token_type_ids': [[0, 0, 0, 0, 0], [0, 0, 0]]}
                    - If False, encoded inputs would be a list like:
                        [{'input_ids': [1, 4444, 4385, 1545, 6712],
                          'token_type_ids': [0, 0, 0, 0, 0]},
                         {'input_ids': [1, 4444, 4385], 'token_type_ids': [0, 0, 0]}]

                Defaults to `True`.
            return_offsets_mapping (bool, optional):
                Whether to include the list of pair preserving the index of start
                and end char in original input for each token in the returned
                dictionary. Would be automatically set to `True` when `stride` > 0.
                Defaults to `False`.
            add_special_tokens (bool, optional):
                Whether to add the special tokens associated with the corresponding model
                to the encoded inputs. Defaults to `True`
            pad_to_multiple_of (int, optional):
                If set will pad the sequence to a multiple of the provided value. This is especially useful to enable
                the use of Tensor Cores on NVIDIA hardware with compute capability >= 7.5 (Volta).
                Defaults to `None`.
            padding_side (`str`, *optional*):
                The side on which the model should have padding applied. Should be selected between ['right', 'left'].
                Default value is picked from the class attribute of the same name.
            return_tensors (str or [TensorType], optional):
                If set, will return tensors instead of list of python integers. Acceptable values are:

                - `'pd'`: Return Paddle `paddle.Tensor` objects.
                - `'np'`: Return Numpy `np.ndarray` objects.
                Defaults to `None`.
            verbose (bool, optional):
                Whether or not to print more information and warnings. Defaults to True.

        Returns:
            dict or list[dict] (for batch input):
                The dict has the following optional items:

                - **input_ids** (list[int] or list[list[int]]): List of token ids to be fed to a model.
                - **position_ids** (list[int] or list[list[int]], optional): List of token position ids to be
                  fed to a model. Included when `return_position_ids` is `True`
                - **token_type_ids** (list[int] or list[list[int]], optional): List of token type ids to be
                  fed to a model. Included when `return_token_type_ids` is `True`.
                - **attention_mask** (list[int] or list[list[int]], optional): List of integers valued 0 or 1,
                  where 0 specifies paddings and should not be attended to by the
                  model. Included when `return_attention_mask` is `True`.
                - **seq_len** (int or list[int], optional): The input_ids length. Included when `return_length`
                  is `True`.
                - **overflowing_tokens** (list[int] or list[list[int]], optional): List of overflowing tokens.
                  Included when if `max_length` is specified and `return_overflowing_tokens`
                  is True.
                - **num_truncated_tokens** (int or list[int], optional): The number of overflowing tokens.
                  Included when if `max_length` is specified and `return_overflowing_tokens`
                  is True.
                - **special_tokens_mask** (list[int] or list[list[int]], optional): List of integers valued 0 or 1,
                  with 0 specifying special added tokens and 1 specifying sequence tokens.
                  Included when `return_special_tokens_mask` is `True`.
                - **offset_mapping** (list[int], optional): list of pair preserving the
                  index of start and end char in original input for each token.
                  For a sqecial token, the index pair is `(0, 0)`. Included when
                  `return_overflowing_tokens` is True or `stride` > 0.
                - **overflow_to_sample** (int or list[int], optional): Index of example from which this
                  feature is generated. Included when `stride` works.
        """

        # Input type checking for clearer error
        def _is_valid_text_input(t):
            if isinstance(t, str):
                # Strings are fine
                return True
            elif isinstance(t, (list, tuple)):
                # List are fine as long as they are...
                if len(t) == 0:
                    # ... empty
                    return True
                elif isinstance(t[0], str):
                    # ... list of strings
                    return True
                elif isinstance(t[0], (list, tuple)):
                    # ... list with an empty list or with a list of strings
                    return len(t[0]) == 0 or isinstance(t[0][0], str)
                else:
                    return False
            else:
                return False

        if not _is_valid_text_input(text):
            raise ValueError(
                "text input must of type `str` (single example), `List[str]` (batch or single pretokenized example) "
                "or `List[List[str]]` (batch of pretokenized examples)."
            )

        if text_pair is not None and not _is_valid_text_input(text_pair):
            raise ValueError(
                "text input must of type `str` (single example), `List[str]` (batch or single pretokenized example) "
                "or `List[List[str]]` (batch of pretokenized examples)."
            )

        # check `split_into_words` value
        if isinstance(is_split_into_words, str) and is_split_into_words != "token":
            raise ValueError(
                "the value of `is_split_into_words` should be one of: {True, False, 'token'} but receive: <%s>",
                is_split_into_words,
            )

        if is_split_into_words:
            is_batched = (
                isinstance(text, (list, tuple))
                and text
                and isinstance(text[0], (list, tuple))
            )
        else:
            is_batched = isinstance(text, (list, tuple))

        if is_batched:
            if isinstance(text_pair, str):
                raise TypeError(
                    "when tokenizing batches of text, `text_pair` must be a list or tuple with the same length as `text`."
                )
            if text_pair is not None and len(text) != len(text_pair):
                raise ValueError(
                    f"batch length of `text`: {len(text)} does not match batch length of `text_pair`: {len(text_pair)}."
                )
            batch_text_or_text_pairs = (
                list(zip(text, text_pair)) if text_pair is not None else text
            )
            return self.batch_encode(
                batch_text_or_text_pairs=batch_text_or_text_pairs,
                max_length=max_length,
                stride=stride,
                is_split_into_words=is_split_into_words,
                padding=padding,
                truncation=truncation,
                return_position_ids=return_position_ids,
                return_token_type_ids=return_token_type_ids,
                return_attention_mask=return_attention_mask,
                return_length=return_length,
                return_overflowing_tokens=return_overflowing_tokens,
                return_special_tokens_mask=return_special_tokens_mask,
                return_dict=return_dict,
                return_offsets_mapping=return_offsets_mapping,
                add_special_tokens=add_special_tokens,
                pad_to_multiple_of=pad_to_multiple_of,
                padding_side=padding_side,
                return_tensors=return_tensors,
                verbose=verbose,
                **kwargs,
            )
        else:
            return self.encode(
                text=text,
                text_pair=text_pair,
                max_length=max_length,
                stride=stride,
                is_split_into_words=is_split_into_words,
                padding=padding,
                truncation=truncation,
                return_position_ids=return_position_ids,
                return_token_type_ids=return_token_type_ids,
                return_attention_mask=return_attention_mask,
                return_length=return_length,
                return_overflowing_tokens=return_overflowing_tokens,
                return_special_tokens_mask=return_special_tokens_mask,
                return_offsets_mapping=return_offsets_mapping,
                add_special_tokens=add_special_tokens,
                pad_to_multiple_of=pad_to_multiple_of,
                padding_side=padding_side,
                return_tensors=return_tensors,
                verbose=verbose,
                **kwargs,
            )

    def encode(
        self,
        text,
        text_pair=None,
        add_special_tokens=True,
        padding: Union[bool, str, PaddingStrategy] = False,
        truncation: Union[bool, str, TruncationStrategy] = False,
        max_length: Optional[int] = None,
        stride: int = 0,
        is_split_into_words: bool = False,
        pad_to_multiple_of: Optional[int] = None,
        padding_side: Optional[Literal["right", "left"]] = None,
        return_tensors: Optional[Union[str, TensorType]] = None,
        return_token_type_ids: Optional[bool] = None,
        return_attention_mask: Optional[bool] = None,
        return_overflowing_tokens: bool = False,
        return_special_tokens_mask: bool = False,
        return_offsets_mapping: bool = False,
        return_length: bool = False,
        verbose: bool = True,
        return_position_ids=None,
        **kwargs,
    ) -> BatchEncoding:
        """
        Tokenize and prepare for the model a sequence or a pair of sequences.

        Args:
            text (`str`, `List[str]` or `List[int]`):
                The first sequence to be encoded. This can be a string, a list of strings (tokenized string using the
                `tokenize` method) or a list of integers (tokenized string ids using the `convert_tokens_to_ids`
                method).
            text_pair (`str`, `List[str]` or `List[int]`, *optional*):
                Optional second sequence to be encoded. This can be a string, a list of strings (tokenized string using
                the `tokenize` method) or a list of integers (tokenized string ids using the `convert_tokens_to_ids`
                method).
        """
        # Backward compatibility for 'max_seq_len'
        old_max_seq_len = kwargs.get("max_seq_len", None)
        if max_length is None and old_max_seq_len:
            if verbose:
                warnings.warn(
                    "The `max_seq_len` argument is deprecated and will be removed in a future version, "
                    "please use `max_length` instead.",
                    FutureWarning,
                )
            max_length = old_max_seq_len
        # Backward compatibility for 'truncation_strategy', 'pad_to_max_length'
        padding_strategy, truncation_strategy, max_length, kwargs = (
            self._get_padding_truncation_strategies(
                padding=padding,
                truncation=truncation,
                max_length=max_length,
                pad_to_multiple_of=pad_to_multiple_of,
                verbose=verbose,
                **kwargs,
            )
        )

        return self._encode_plus(
            text=text,
            text_pair=text_pair,
            add_special_tokens=add_special_tokens,
            padding_strategy=padding_strategy,
            truncation_strategy=truncation_strategy,
            max_length=max_length,
            stride=stride,
            is_split_into_words=is_split_into_words,
            pad_to_multiple_of=pad_to_multiple_of,
            padding_side=padding_side,
            return_tensors=return_tensors,
            return_position_ids=return_position_ids,
            return_token_type_ids=return_token_type_ids,
            return_attention_mask=return_attention_mask,
            return_overflowing_tokens=return_overflowing_tokens,
            return_special_tokens_mask=return_special_tokens_mask,
            return_offsets_mapping=return_offsets_mapping,
            return_length=return_length,
            verbose=verbose,
            **kwargs,
        )

    def encode_plus(
        self,
        text: Union[TextInput, PreTokenizedInput, EncodedInput],
        text_pair: Optional[Union[TextInput, PreTokenizedInput, EncodedInput]] = None,
        add_special_tokens: bool = True,
        padding: Union[bool, str, PaddingStrategy] = False,
        truncation: Union[bool, str, TruncationStrategy] = None,
        max_length: Optional[int] = None,
        stride: int = 0,
        is_split_into_words: bool = False,
        padding_side: Optional[Literal["right", "left"]] = None,
        pad_to_multiple_of: Optional[int] = None,
        return_tensors: Optional[Union[str, TensorType]] = None,
        return_token_type_ids: Optional[bool] = None,
        return_attention_mask: Optional[bool] = None,
        return_overflowing_tokens: bool = False,
        return_special_tokens_mask: bool = False,
        return_offsets_mapping: bool = False,
        return_length: bool = False,
        verbose: bool = True,
        **kwargs,
    ) -> BatchEncoding:
        """
        Tokenize and prepare for the model a sequence or a pair of sequences.

        <Tip warning={true}>

        This method is deprecated, `__call__` should be used instead.

        </Tip>

        Args:
            text (`str`, `List[str]` or `List[int]` (the latter only for not-fast tokenizers)):
                The first sequence to be encoded. This can be a string, a list of strings (tokenized string using the
                `tokenize` method) or a list of integers (tokenized string ids using the `convert_tokens_to_ids`
                method).
            text_pair (`str`, `List[str]` or `List[int]`, *optional*):
                Optional second sequence to be encoded. This can be a string, a list of strings (tokenized string using
                the `tokenize` method) or a list of integers (tokenized string ids using the `convert_tokens_to_ids`
                method).
        """

        # Backward compatibility for 'truncation_strategy', 'pad_to_max_length'
        padding_strategy, truncation_strategy, max_length, kwargs = (
            self._get_padding_truncation_strategies(
                padding=padding,
                truncation=truncation,
                max_length=max_length,
                pad_to_multiple_of=pad_to_multiple_of,
                verbose=verbose,
                **kwargs,
            )
        )

        return self._encode_plus(
            text=text,
            text_pair=text_pair,
            add_special_tokens=add_special_tokens,
            padding_strategy=padding_strategy,
            truncation_strategy=truncation_strategy,
            max_length=max_length,
            stride=stride,
            is_split_into_words=is_split_into_words,
            pad_to_multiple_of=pad_to_multiple_of,
            padding_side=padding_side,
            return_tensors=return_tensors,
            return_token_type_ids=return_token_type_ids,
            return_attention_mask=return_attention_mask,
            return_overflowing_tokens=return_overflowing_tokens,
            return_special_tokens_mask=return_special_tokens_mask,
            return_offsets_mapping=return_offsets_mapping,
            return_length=return_length,
            verbose=verbose,
            **kwargs,
        )

    def _encode_plus(
        self,
        text: Union[TextInput, PreTokenizedInput, EncodedInput],
        text_pair: Optional[Union[TextInput, PreTokenizedInput, EncodedInput]] = None,
        add_special_tokens: bool = True,
        padding_strategy: PaddingStrategy = PaddingStrategy.DO_NOT_PAD,
        truncation_strategy: TruncationStrategy = TruncationStrategy.DO_NOT_TRUNCATE,
        max_length: Optional[int] = None,
        stride: int = 0,
        is_split_into_words: bool = False,
        pad_to_multiple_of: Optional[int] = None,
        padding_side: Optional[Literal["right", "left"]] = None,
        return_position_ids: Optional[bool] = None,
        return_tensors: Optional[Union[str, TensorType]] = None,
        return_token_type_ids: Optional[bool] = None,
        return_attention_mask: Optional[bool] = None,
        return_overflowing_tokens: bool = False,
        return_special_tokens_mask: bool = False,
        return_offsets_mapping: bool = False,
        return_length: bool = False,
        verbose: bool = True,
        **kwargs,
    ) -> BatchEncoding:
        raise NotImplementedError

    def batch_encode(
        self,
        batch_text_or_text_pairs: Union[
            List[TextInput],
            List[TextInputPair],
            List[PreTokenizedInput],
            List[PreTokenizedInputPair],
            List[EncodedInput],
            List[EncodedInputPair],
        ],
        max_length=None,
        stride: int = 0,
        is_split_into_words: bool = False,
        padding: Union[bool, str, PaddingStrategy] = False,
        truncation: Union[bool, str, TruncationStrategy] = False,
        return_position_ids=None,
        # TODO(wj-mcat): keep align with `encode` method
        return_token_type_ids=None,
        return_attention_mask=None,
        return_length=False,
        return_overflowing_tokens=False,
        return_special_tokens_mask=False,
        return_dict=True,
        return_offsets_mapping=False,
        add_special_tokens=True,
        pad_to_multiple_of: Optional[int] = None,
        padding_side: Optional[Literal["right", "left"]] = None,
        return_tensors: Optional[Union[str, TensorType]] = None,
        verbose: bool = True,
        **kwargs,
    ) -> BatchEncoding:
        """
        Performs tokenization and uses the tokenized tokens to prepare model
        inputs. It supports batch inputs of sequence or sequence pair.

        Args:
            batch_text_or_text_pairs (list):
                The element of list can be sequence or sequence pair, and the
                sequence is a string or a list of strings depending on whether
                it has been pretokenized. If each sequence is provided as a list
                of strings (pretokenized), you must set `is_split_into_words` as
                `True` to disambiguate with a sequence pair.

        Returns:
            dict or list[dict]:
                The dict has the following optional items:

        """
        # Backward compatibility for 'max_seq_len'
        old_max_seq_len = kwargs.get("max_seq_len", None)
        if max_length is None and old_max_seq_len:
            if verbose:
                warnings.warn(
                    "The `max_seq_len` argument is deprecated and will be removed in a future version, "
                    "please use `max_length` instead.",
                    FutureWarning,
                )
            max_length = old_max_seq_len
        # Backward compatibility for 'truncation_strategy', 'pad_to_max_length'
        padding_strategy, truncation_strategy, max_length, kwargs = (
            self._get_padding_truncation_strategies(
                padding=padding,
                truncation=truncation,
                max_length=max_length,
                pad_to_multiple_of=pad_to_multiple_of,
                verbose=verbose,
                **kwargs,
            )
        )

        return self._batch_encode_plus(
            batch_text_or_text_pairs=batch_text_or_text_pairs,
            add_special_tokens=add_special_tokens,
            padding_strategy=padding_strategy,
            truncation_strategy=truncation_strategy,
            max_length=max_length,
            stride=stride,
            is_split_into_words=is_split_into_words,
            pad_to_multiple_of=pad_to_multiple_of,
            padding_side=padding_side,
            return_tensors=return_tensors,
            return_position_ids=return_position_ids,
            return_token_type_ids=return_token_type_ids,
            return_attention_mask=return_attention_mask,
            return_overflowing_tokens=return_overflowing_tokens,
            return_special_tokens_mask=return_special_tokens_mask,
            return_dict=return_dict,
            return_offsets_mapping=return_offsets_mapping,
            return_length=return_length,
            verbose=verbose,
            **kwargs,
        )

    def _batch_encode_plus(
        self,
        batch_text_or_text_pairs: Union[
            List[TextInput],
            List[TextInputPair],
            List[PreTokenizedInput],
            List[PreTokenizedInputPair],
            List[EncodedInput],
            List[EncodedInputPair],
        ],
        add_special_tokens: bool = True,
        padding_strategy: PaddingStrategy = PaddingStrategy.DO_NOT_PAD,
        truncation_strategy: TruncationStrategy = TruncationStrategy.DO_NOT_TRUNCATE,
        max_length: Optional[int] = None,
        stride: int = 0,
        is_split_into_words: bool = False,
        pad_to_multiple_of: Optional[int] = None,
        padding_side: Optional[Literal["right", "left"]] = None,
        return_position_ids: Optional[bool] = None,
        return_tensors: Optional[Union[str, TensorType]] = None,
        return_token_type_ids: Optional[bool] = None,
        return_attention_mask: Optional[bool] = None,
        return_overflowing_tokens: bool = False,
        return_special_tokens_mask: bool = False,
        return_dict: bool = True,
        return_offsets_mapping: bool = False,
        return_length: bool = False,
        verbose: bool = True,
        **kwargs,
    ) -> BatchEncoding:
        raise NotImplementedError

    def pad(
        self,
        encoded_inputs: Union[
            BatchEncoding,
            List[BatchEncoding],
            Dict[str, EncodedInput],
            Dict[str, List[EncodedInput]],
            List[Dict[str, EncodedInput]],
        ],
        padding: Union[bool, str, PaddingStrategy] = True,
        max_length: Optional[int] = None,
        padding_side: Optional[Literal["right", "left"]] = None,
        pad_to_multiple_of: Optional[int] = None,
        return_attention_mask: Optional[bool] = None,
        return_tensors: Optional[Union[str, TensorType]] = None,
        verbose: bool = True,
    ) -> BatchEncoding:
        """
        Pad a single encoded input or a batch of encoded inputs up to predefined length or to the max sequence length
        in the batch.

        Padding side (left/right) padding token ids are defined at the tokenizer level (with `self.padding_side`,
        `self.pad_token_id` and `self.pad_token_type_id`)

        <Tip>

        If the `encoded_inputs` passed are dictionary of numpy arrays, Paddle tensors, the
        result will use the same type unless you provide a different tensor type with `return_tensors`.
        </Tip>

        Args:
            encoded_inputs ([`BatchEncoding`], list of [`BatchEncoding`], `Dict[str, List[int]]`, `Dict[str, List[List[int]]` or `List[Dict[str, List[int]]]`):
                Tokenized inputs. Can represent one input ([`BatchEncoding`] or `Dict[str, List[int]]`) or a batch of
                tokenized inputs (list of [`BatchEncoding`], *Dict[str, List[List[int]]]* or *List[Dict[str,
                List[int]]]*) so you can use this method during preprocessing as well as in a Paddle Dataloader
                collate function.

                Instead of `List[int]` you can have tensors (numpy arrays, Paddle tensors), see
                the note above for the return type.
            padding (`bool`, `str` or [`PaddingStrategy`], *optional*, defaults to `True`):
                 Select a strategy to pad the returned sequences (according to the model's padding side and padding
                 index) among:

                - `True` or `'longest'`: Pad to the longest sequence in the batch (or no padding if only a single
                  sequence if provided).
                - `'max_length'`: Pad to a maximum length specified with the argument `max_length` or to the maximum
                  acceptable input length for the model if that argument is not provided.
                - `False` or `'do_not_pad'` (default): No padding (i.e., can output a batch with sequences of different
                  lengths).
            max_length (`int`, *optional*):
                Maximum length of the returned list and optionally padding length (see above).
            pad_to_multiple_of (`int`, *optional*):
                If set will pad the sequence to a multiple of the provided value.

                This is especially useful to enable the use of Tensor Cores on NVIDIA hardware with compute capability
                >= 7.5 (Volta).
            padding_side (`str`, *optional*):
                The side on which the model should have padding applied. Should be selected between ['right', 'left'].
                Default value is picked from the class attribute of the same name.
            return_attention_mask (`bool`, *optional*):
                Whether to return the attention mask. If left to the default, will return the attention mask according
                to the specific tokenizer's default, defined by the `return_outputs` attribute.

                [What are attention masks?](../glossary#attention-mask)
            return_tensors (`str` or [`TensorType`], *optional*):
                If set, will return tensors instead of list of python integers. Acceptable values are:

                - `'pd'`: Return Paddle `paddle.Tensor` objects.
                - `'np'`: Return Numpy `np.ndarray` objects.
            verbose (`bool`, *optional*, defaults to `True`):
                Whether or not to print more information and warnings.
        """
        import paddle

        # If we have a list of dicts, let's convert it in a dict of lists
        if isinstance(encoded_inputs, (list, tuple)) and isinstance(
            encoded_inputs[0], (dict, BatchEncoding)
        ):
            encoded_inputs = {
                key: [example[key] for example in encoded_inputs]
                for key in encoded_inputs[0].keys()
            }

        # The model's main input name, usually `input_ids`, has be passed for padding
        if self.model_input_names[0] not in encoded_inputs:
            raise ValueError(
                "You should supply an encoding or a list of encodings to this method "
                f"that includes {self.model_input_names[0]}, but you provided {list(encoded_inputs.keys())}"
            )

        required_input = encoded_inputs[self.model_input_names[0]]

        if not required_input:
            if return_attention_mask:
                encoded_inputs["attention_mask"] = []
            return encoded_inputs

        # If we have Paddle/NumPy tensors/arrays as inputs, we cast them as python objects
        # and rebuild them afterwards if no return_tensors is specified

        first_element = required_input[0]
        if isinstance(first_element, (list, tuple)):
            # first_element might be an empty list/tuple in some edge cases so we grab the first non empty element.
            for item in required_input:
                if len(item) != 0:
                    first_element = item[0]
                    break
        # At this state, if `first_element` is still a list/tuple, it's an empty one so there is nothing to do.
        if not isinstance(first_element, (int, list, tuple)):
            if isinstance(first_element, paddle.Tensor):
                return_tensors = "pd" if return_tensors is None else return_tensors
            else:
                raise ValueError(
                    f"type of {first_element} unknown: {type(first_element)}. "
                    f"Should be either python or paddle object."
                )

            for key, value in encoded_inputs.items():
                encoded_inputs[key] = to_py_obj(value)

        # Convert padding_strategy in PaddingStrategy
        padding_strategy, _, max_length, _ = self._get_padding_truncation_strategies(
            padding=padding, max_length=max_length, verbose=verbose
        )

        required_input = encoded_inputs[self.model_input_names[0]]
        if required_input and not isinstance(required_input[0], (list, tuple)):
            # some tokenizers might not have the padding_side attribute
            if "padding_side" in set(inspect.signature(self._pad).parameters.keys()):
                encoded_inputs = self._pad(
                    encoded_inputs,
                    max_length=max_length,
                    padding_strategy=padding_strategy,
                    pad_to_multiple_of=pad_to_multiple_of,
                    padding_side=padding_side,
                    return_attention_mask=return_attention_mask,
                )
            else:
                original_padding_side = self.padding_side
                self.padding_side = padding_side
                encoded_inputs = self._pad(
                    encoded_inputs,
                    max_length=max_length,
                    padding_strategy=padding_strategy,
                    pad_to_multiple_of=pad_to_multiple_of,
                    return_attention_mask=return_attention_mask,
                )
                self.padding_side = original_padding_side

            return BatchEncoding(encoded_inputs, tensor_type=return_tensors)

        batch_size = len(required_input)
        assert all(
            len(v) == batch_size for v in encoded_inputs.values()
        ), "Some items in the output dictionary have a different batch size than others."

        if padding_strategy == PaddingStrategy.LONGEST:
            max_length = max(len(inputs) for inputs in required_input)
            padding_strategy = PaddingStrategy.MAX_LENGTH

        batch_outputs = {}
        for i in range(batch_size):
            inputs = dict((k, v[i]) for k, v in encoded_inputs.items())
            outputs = self._pad(
                inputs,
                max_length=max_length,
                padding_strategy=padding_strategy,
                padding_side=padding_side,
                pad_to_multiple_of=pad_to_multiple_of,
                return_attention_mask=return_attention_mask,
            )

            for key, value in outputs.items():
                if key not in batch_outputs:
                    batch_outputs[key] = []
                batch_outputs[key].append(value)

        return BatchEncoding(batch_outputs, tensor_type=return_tensors)

    def create_token_type_ids_from_sequences(
        self, token_ids_0: List[int], token_ids_1: Optional[List[int]] = None
    ) -> List[int]:
        """
        Create the token type IDs corresponding to the sequences passed. [What are token type
        IDs?](../glossary#token-type-ids)

        Should be overridden in a subclass if the model has a special way of building those.

        Args:
            token_ids_0 (`List[int]`): The first tokenized sequence.
            token_ids_1 (`List[int]`, *optional*): The second tokenized sequence.

        Returns:
            `List[int]`: The token type ids.
        """
        if token_ids_1 is None:
            return len(token_ids_0) * [0]
        return [0] * len(token_ids_0) + [1] * len(token_ids_1)

    def build_inputs_with_special_tokens(
        self, token_ids_0: List[int], token_ids_1: Optional[List[int]] = None
    ) -> List[int]:
        """
        Build model inputs from a sequence or a pair of sequence for sequence classification tasks by concatenating and
        adding special tokens.

        This implementation does not add special tokens and this method should be overridden in a subclass.

        Args:
            token_ids_0 (`List[int]`): The first tokenized sequence.
            token_ids_1 (`List[int]`, *optional*): The second tokenized sequence.

        Returns:
            `List[int]`: The model input with special tokens.
        """
        if token_ids_1 is None:
            return token_ids_0
        return token_ids_0 + token_ids_1

    def build_offset_mapping_with_special_tokens(
        self, offset_mapping_0, offset_mapping_1=None
    ):
        """
        Build offset map from a pair of offset map by concatenating and adding offsets of special tokens.

        Should be overridden in a subclass if the model has a special way of building those.

        Args:
            offset_mapping_0 (List[tuple]):
                List of char offsets to which the special tokens will be added.
            offset_mapping_1 (List[tuple], optional):
                Optional second list of char offsets for offset mapping pairs.

        Returns:
            List[tuple]: List of char offsets with the appropriate offsets of special tokens.
        """
        if offset_mapping_1 is None:
            return offset_mapping_0

        return offset_mapping_0 + offset_mapping_1

    def prepare_for_model(
        self,
        ids,
        pair_ids=None,
        padding: Union[bool, str, PaddingStrategy] = False,
        truncation: Union[bool, str, TruncationStrategy] = False,
        max_length: Optional[int] = None,
        stride: int = 0,
        pad_to_multiple_of: Optional[int] = None,
        padding_side: Optional[Literal["right", "left"]] = None,
        return_tensors: Optional[Union[str, TensorType]] = None,
        return_position_ids=None,
        return_token_type_ids: Optional[bool] = None,
        return_attention_mask: Optional[bool] = None,
        return_length=False,
        return_overflowing_tokens=False,
        return_special_tokens_mask=False,
        return_offsets_mapping=False,
        add_special_tokens=True,
        verbose: bool = True,
        prepend_batch_axis: bool = False,
        **kwargs,
    ):
        """
        Performs tokenization and uses the tokenized tokens to prepare model
        inputs. It supports sequence or sequence pair as input, and batch input
        is not allowed.
        """
        padding_strategy, truncation_strategy, max_length, kwargs = (
            self._get_padding_truncation_strategies(
                padding=padding,
                truncation=truncation,
                max_length=max_length,
                pad_to_multiple_of=pad_to_multiple_of,
                verbose=verbose,
                **kwargs,
            )
        )

        pair = bool(pair_ids is not None)
        len_ids = len(ids)
        len_pair_ids = len(pair_ids) if pair else 0

        if return_token_type_ids and not add_special_tokens:
            raise ValueError(
                "Asking to return token_type_ids while setting add_special_tokens to False "
                "results in an undefined behavior. Please set add_special_tokens to True or "
                "set return_token_type_ids to None."
            )

        if (
            return_overflowing_tokens
            and truncation_strategy == TruncationStrategy.LONGEST_FIRST
            and pair_ids is not None
        ):
            raise ValueError(
                "Not possible to return overflowing tokens for pair of sequences with the "
                "`longest_first`. Please select another truncation strategy than `longest_first`, "
                "for instance `only_second` or `only_first`."
            )

        # Load from model defaults
        if return_token_type_ids is None:
            return_token_type_ids = "token_type_ids" in self.model_input_names
        if return_attention_mask is None:
            return_attention_mask = "attention_mask" in self.model_input_names
        if return_position_ids is None:
            return_position_ids = "position_ids" in self.model_input_names
        encoded_inputs = {}
        # Truncation: Handle max sequence length
        total_len = (
            len_ids
            + len_pair_ids
            + (self.num_special_tokens_to_add(pair=pair) if add_special_tokens else 0)
        )

        overflowing_tokens = []

        if (
            truncation_strategy != TruncationStrategy.DO_NOT_TRUNCATE
            and max_length
            and total_len > max_length
        ):
            ids, pair_ids, overflowing_tokens = self.truncate_sequences(
                ids,
                pair_ids=pair_ids,
                num_tokens_to_remove=total_len - max_length,
                truncation_strategy=truncation_strategy,
                stride=stride,
            )
        if return_overflowing_tokens:
            encoded_inputs["overflowing_tokens"] = overflowing_tokens
            encoded_inputs["num_truncated_tokens"] = total_len - max_length

        # Add special tokens
        if add_special_tokens:
            sequence = self.build_inputs_with_special_tokens(ids, pair_ids)
            token_type_ids = self.create_token_type_ids_from_sequences(ids, pair_ids)
        else:
            sequence = ids + pair_ids if pair else ids
            token_type_ids = [0] * len(ids) + ([0] * len(pair_ids) if pair else [])

        # Build output dictionary
        encoded_inputs["input_ids"] = sequence
        if return_token_type_ids:
            encoded_inputs["token_type_ids"] = token_type_ids
        if return_special_tokens_mask:
            if add_special_tokens:
                encoded_inputs["special_tokens_mask"] = self.get_special_tokens_mask(
                    ids, pair_ids
                )
            else:
                encoded_inputs["special_tokens_mask"] = [0] * len(sequence)

        if return_offsets_mapping and "text" in kwargs and "text_pair" in kwargs:
            text = kwargs.pop("text")
            text_pair = kwargs.pop("text_pair")

            token_offset_mapping = self.get_offset_mapping(text)
            token_pair_offset_mapping = (
                self.get_offset_mapping(text_pair) if text_pair is not None else None
            )
            if max_length and total_len > max_length:
                token_offset_mapping, token_pair_offset_mapping, _ = (
                    self.truncate_sequences(
                        token_offset_mapping,
                        pair_ids=token_pair_offset_mapping,
                        num_tokens_to_remove=total_len - max_length,
                        truncation_strategy=truncation_strategy,
                        stride=stride,
                    )
                )
            if add_special_tokens:
                offset_mapping = self.build_offset_mapping_with_special_tokens(
                    token_offset_mapping, token_pair_offset_mapping
                )
            else:
                offset_mapping = (
                    token_offset_mapping + token_pair_offset_mapping
                    if token_pair_offset_mapping
                    else token_offset_mapping
                )
            encoded_inputs["offset_mapping"] = offset_mapping

        # Check lengths
        self._eventual_warn_about_too_long_sequence(
            encoded_inputs["input_ids"], max_length, verbose
        )

        if return_position_ids:
            encoded_inputs["position_ids"] = list(
                range(len(encoded_inputs["input_ids"]))
            )

        if padding_strategy != PaddingStrategy.DO_NOT_PAD or return_attention_mask:
            encoded_inputs = self.pad(
                encoded_inputs,
                max_length=max_length,
                padding=padding_strategy.value,
                pad_to_multiple_of=pad_to_multiple_of,
                padding_side=padding_side,
                return_attention_mask=return_attention_mask,
            )

        if return_length:
            encoded_inputs["length"] = len(encoded_inputs["input_ids"])
            # for compatibility
            encoded_inputs["seq_len"] = encoded_inputs["length"]

        batch_outputs = BatchEncoding(
            encoded_inputs,
            tensor_type=return_tensors,
            prepend_batch_axis=prepend_batch_axis,
        )

        return batch_outputs

    def truncate_sequences(
        self,
        ids: List[int],
        pair_ids: Optional[List[int]] = None,
        num_tokens_to_remove: int = 0,
        truncation_strategy: Union[str, TruncationStrategy] = "longest_first",
        stride: int = 0,
    ) -> Tuple[List[int], List[int], List[int]]:
        """
        Truncates a sequence pair in-place following the strategy.

        Args:
            ids (`List[int]`):
                Tokenized input ids of the first sequence. Can be obtained from a string by chaining the `tokenize` and
                `convert_tokens_to_ids` methods.
            pair_ids (`List[int]`, *optional*):
                Tokenized input ids of the second sequence. Can be obtained from a string by chaining the `tokenize`
                and `convert_tokens_to_ids` methods.
            num_tokens_to_remove (`int`, *optional*, defaults to 0):
                Number of tokens to remove using the truncation strategy.
            truncation_strategy (`str` or [`TruncationStrategy`], *optional*, defaults to `False`):
                The strategy to follow for truncation. Can be:

                - `'longest_first'`: Truncate to a maximum length specified with the argument `max_length` or to the
                  maximum acceptable input length for the model if that argument is not provided. This will truncate
                  token by token, removing a token from the longest sequence in the pair if a pair of sequences (or a
                  batch of pairs) is provided.
                - `'only_first'`: Truncate to a maximum length specified with the argument `max_length` or to the
                  maximum acceptable input length for the model if that argument is not provided. This will only
                  truncate the first sequence of a pair if a pair of sequences (or a batch of pairs) is provided.
                - `'only_second'`: Truncate to a maximum length specified with the argument `max_length` or to the
                  maximum acceptable input length for the model if that argument is not provided. This will only
                  truncate the second sequence of a pair if a pair of sequences (or a batch of pairs) is provided.
                - `'do_not_truncate'` (default): No truncation (i.e., can output batch with sequence lengths greater
                  than the model maximum admissible input size).
            stride (`int`, *optional*, defaults to 0):
                If set to a positive number, the overflowing tokens returned will contain some tokens from the main
                sequence returned. The value of this argument defines the number of additional tokens.

        Returns:
            `Tuple[List[int], List[int], List[int]]`: The truncated `ids`, the truncated `pair_ids` and the list of
            overflowing tokens. Note: The *longest_first* strategy returns empty list of overflowing tokens if a pair
            of sequences (or a batch of pairs) is provided.
        """
        if num_tokens_to_remove <= 0:
            return ids, pair_ids, []

        if not isinstance(truncation_strategy, TruncationStrategy):
            truncation_strategy = TruncationStrategy(truncation_strategy)

        overflowing_tokens = []
        if truncation_strategy == TruncationStrategy.ONLY_FIRST or (
            truncation_strategy == TruncationStrategy.LONGEST_FIRST and pair_ids is None
        ):
            if len(ids) > num_tokens_to_remove:
                window_len = min(len(ids), stride + num_tokens_to_remove)
                if self.truncation_side == "left":
                    overflowing_tokens = ids[:window_len]
                    ids = ids[num_tokens_to_remove:]
                elif self.truncation_side == "right":
                    overflowing_tokens = ids[-window_len:]
                    ids = ids[:-num_tokens_to_remove]
                else:
                    raise ValueError(
                        f"invalid truncation strategy: {self.truncation_side}, use 'left' or 'right'."
                    )

            else:
                error_msg = (
                    f"We need to remove {num_tokens_to_remove} to truncate the input "
                    f"but the first sequence has a length {len(ids)}. "
                )
                if truncation_strategy == TruncationStrategy.ONLY_FIRST:
                    error_msg = (
                        error_msg + "Please select another truncation strategy than "
                        f"{truncation_strategy}, for instance 'longest_first' or 'only_second'."
                    )
                logging.error(error_msg)
        elif truncation_strategy == TruncationStrategy.LONGEST_FIRST:
            warnings.warn(
                f"Be aware, overflowing tokens are not returned for the setting you have chosen,"
                f" i.e. sequence pairs with the '{TruncationStrategy.LONGEST_FIRST.value}' "
                f"truncation strategy. So the returned list will always be empty even if some "
                f"tokens have been removed."
            )
            for _ in range(num_tokens_to_remove):
                if pair_ids is None or len(ids) > len(pair_ids):
                    if self.truncation_side == "right":
                        ids = ids[:-1]
                    elif self.truncation_side == "left":
                        ids = ids[1:]
                    else:
                        raise ValueError(
                            "invalid truncation strategy:" + str(self.truncation_side)
                        )
                else:
                    if self.truncation_side == "right":
                        pair_ids = pair_ids[:-1]
                    elif self.truncation_side == "left":
                        pair_ids = pair_ids[1:]
                    else:
                        raise ValueError(
                            "invalid truncation strategy:" + str(self.truncation_side)
                        )
        elif (
            truncation_strategy == TruncationStrategy.ONLY_SECOND
            and pair_ids is not None
        ):
            if len(pair_ids) > num_tokens_to_remove:
                window_len = min(len(pair_ids), stride + num_tokens_to_remove)
                if self.truncation_side == "right":
                    overflowing_tokens = pair_ids[-window_len:]
                    pair_ids = pair_ids[:-num_tokens_to_remove]
                elif self.truncation_side == "left":
                    overflowing_tokens = pair_ids[:window_len]
                    pair_ids = pair_ids[num_tokens_to_remove:]
                else:
                    raise ValueError(
                        "invalid truncation strategy:" + str(self.truncation_side)
                    )
            else:
                logging.error(
                    f"We need to remove {num_tokens_to_remove} to truncate the input "
                    f"but the second sequence has a length {len(pair_ids)}. "
                    f"Please select another truncation strategy than {truncation_strategy}, "
                    f"for instance 'longest_first' or 'only_first'."
                )

        return (ids, pair_ids, overflowing_tokens)

    def _pad(
        self,
        encoded_inputs: Union[Dict[str, EncodedInput], BatchEncoding],
        max_length: Optional[int] = None,
        padding_strategy: PaddingStrategy = PaddingStrategy.DO_NOT_PAD,
        pad_to_multiple_of: Optional[int] = None,
        padding_side: Optional[Literal["right", "left"]] = None,
        return_attention_mask: Optional[bool] = None,
    ) -> dict:
        """
        Pad encoded inputs (on left/right and up to predefined length or max length in the batch)

        Args:
            encoded_inputs:
                Dictionary of tokenized inputs (`List[int]`) or batch of tokenized inputs (`List[List[int]]`).
            max_length: maximum length of the returned list and optionally padding length (see below).
                Will truncate by taking into account the special tokens.
            padding_strategy: PaddingStrategy to use for padding.

                - PaddingStrategy.LONGEST Pad to the longest sequence in the batch
                - PaddingStrategy.MAX_LENGTH: Pad to the max length (default)
                - PaddingStrategy.DO_NOT_PAD: Do not pad
                The tokenizer padding sides are defined in `padding_side` argument:

                    - 'left': pads on the left of the sequences
                    - 'right': pads on the right of the sequences
            pad_to_multiple_of: (optional) Integer if set will pad the sequence to a multiple of the provided value.
                This is especially useful to enable the use of Tensor Core on NVIDIA hardware with compute capability
                >= 7.5 (Volta).
            padding_side: (optional) The side on which the model should have padding applied.
                Should be selected between ['right', 'left'].
                Default value is picked from the class attribute of the same name.
            return_attention_mask:
                (optional) Set to False to avoid returning attention mask (default: set to model specifics)
        """
        # Load from model defaults
        if return_attention_mask is None:
            return_attention_mask = (
                "attention_mask" in self.model_input_names
                or "attention_mask" in encoded_inputs
            )

        required_input = encoded_inputs[self.model_input_names[0]]

        if padding_strategy == PaddingStrategy.LONGEST:
            max_length = len(required_input)

        if (
            max_length is not None
            and pad_to_multiple_of is not None
            and (max_length % pad_to_multiple_of != 0)
        ):
            max_length = ((max_length // pad_to_multiple_of) + 1) * pad_to_multiple_of

        needs_to_be_padded = (
            padding_strategy != PaddingStrategy.DO_NOT_PAD
            and len(required_input) != max_length
        )

        # Initialize attention mask if not present.
        if return_attention_mask and "attention_mask" not in encoded_inputs:
            encoded_inputs["attention_mask"] = [1] * len(required_input)

        if needs_to_be_padded:
            difference = max_length - len(required_input)
            padding_side = (
                padding_side if padding_side is not None else self.padding_side
            )

            if padding_side == "right":
                if return_attention_mask:
                    if len(np.shape(encoded_inputs["attention_mask"])) > 2:
                        encoded_inputs["attention_mask"] = np.pad(
                            encoded_inputs["attention_mask"],
                            pad_width=[(0, 0), (0, difference), (0, difference)],
                            mode="constant",
                            constant_values=0,
                        ).tolist()
                    else:
                        encoded_inputs["attention_mask"] = (
                            encoded_inputs["attention_mask"] + [0] * difference
                        )
                if "attn_mask_startend_row_indices" in encoded_inputs:
                    encoded_inputs["attn_mask_startend_row_indices"] = np.concatenate(
                        [
                            np.array(
                                [encoded_inputs["attn_mask_startend_row_indices"]],
                                dtype=np.int32,
                            ),
                            np.zeros([1, difference], dtype=np.int32),
                        ],
                        axis=-1,
                    )
                if "token_type_ids" in encoded_inputs:
                    encoded_inputs["token_type_ids"] = (
                        encoded_inputs["token_type_ids"]
                        + [self.pad_token_type_id] * difference
                    )
                if "special_tokens_mask" in encoded_inputs:
                    encoded_inputs["special_tokens_mask"] = (
                        encoded_inputs["special_tokens_mask"] + [1] * difference
                    )
                if "offset_mapping" in encoded_inputs:
                    encoded_inputs["offset_mapping"] = (
                        encoded_inputs["offset_mapping"] + [(0, 0)] * difference
                    )
                if "position_ids" in encoded_inputs:
                    encoded_inputs["position_ids"] = (
                        encoded_inputs["position_ids"] + [0] * difference
                    )
                # NOTE: In ernie3.0-qa, the type of `*_positions` is int.
                if "start_positions" in encoded_inputs and isinstance(
                    encoded_inputs["start_positions"], list
                ):
                    encoded_inputs["start_positions"] = (
                        encoded_inputs["start_positions"] + [0] * difference
                    )
                if "end_positions" in encoded_inputs and isinstance(
                    encoded_inputs["end_positions"], list
                ):
                    encoded_inputs["end_positions"] = (
                        encoded_inputs["end_positions"] + [0] * difference
                    )
                encoded_inputs[self.model_input_names[0]] = (
                    required_input + [self.pad_token_id] * difference
                )
            elif padding_side == "left":
                if return_attention_mask:
                    if len(np.shape(encoded_inputs["attention_mask"])) > 2:
                        # attention_mask shape [1,seq_len,seq_len]
                        encoded_inputs["attention_mask"] = np.pad(
                            encoded_inputs["attention_mask"],
                            pad_width=[(0, 0), (difference, 0), (difference, 0)],
                            mode="constant",
                            constant_values=0,
                        ).tolist()
                    else:
                        encoded_inputs["attention_mask"] = [
                            0
                        ] * difference + encoded_inputs["attention_mask"]
                if "attn_mask_startend_row_indices" in encoded_inputs:
                    encoded_inputs["attn_mask_startend_row_indices"] = np.concatenate(
                        [
                            np.zeros([1, difference], dtype=np.int32),
                            np.array(
                                [encoded_inputs["attn_mask_startend_row_indices"]],
                                dtype=np.int32,
                            )
                            + difference,
                        ],
                        axis=-1,
                    )
                if "token_type_ids" in encoded_inputs:
                    encoded_inputs["token_type_ids"] = [
                        self.pad_token_type_id
                    ] * difference + encoded_inputs["token_type_ids"]
                if "special_tokens_mask" in encoded_inputs:
                    encoded_inputs["special_tokens_mask"] = [
                        1
                    ] * difference + encoded_inputs["special_tokens_mask"]
                if "offset_mapping" in encoded_inputs:
                    encoded_inputs["offset_mapping"] = [
                        (0, 0)
                    ] * difference + encoded_inputs["offset_mapping"]
                if "position_ids" in encoded_inputs:
                    encoded_inputs["position_ids"] = [0] * difference + encoded_inputs[
                        "position_ids"
                    ]
                if "start_positions" in encoded_inputs and isinstance(
                    encoded_inputs["start_positions"], list
                ):
                    encoded_inputs["start_positions"] = [
                        0
                    ] * difference + encoded_inputs["start_positions"]
                if "end_positions" in encoded_inputs and isinstance(
                    encoded_inputs["end_positions"], list
                ):
                    encoded_inputs["end_positions"] = [0] * difference + encoded_inputs[
                        "end_positions"
                    ]
                encoded_inputs[self.model_input_names[0]] = [
                    self.pad_token_id
                ] * difference + required_input
            else:
                raise ValueError("Invalid padding strategy:" + str(self.padding_side))
        else:
            if "attn_mask_startend_row_indices" in encoded_inputs:
                if len(np.shape(encoded_inputs["attn_mask_startend_row_indices"])) == 1:
                    encoded_inputs["attn_mask_startend_row_indices"] = np.array([encoded_inputs["attn_mask_startend_row_indices"]], dtype=np.int32)  # fmt:skip

        if "attn_mask_startend_row_indices" in encoded_inputs:
            assert (
                len(np.shape(encoded_inputs["attn_mask_startend_row_indices"])) == 2
            )  # [num_head, seq_len]

        return encoded_inputs

    def convert_tokens_to_string(self, tokens: List[str]) -> str:
        """
        Converts a sequence of tokens in a single string. The most simple way to do it is `" ".join(tokens)` but we
        often want to remove sub-word tokenization artifacts at the same time.

        Args:
            tokens (`List[str]`): The token to join in a string.

        Returns:
            `str`: The joined tokens.
        """
        raise NotImplementedError

    def decode_token(
        self,
        all_input_ids: List[int],
        prefix_offset: int = 0,
        read_offset: int = 0,
    ) -> Tuple[str, int, int]:
        """tokenizer decoding for the streaming generation use case. This method can be overridden for tokenizer that doesn't follow this API"""
        prefix_text = self.decode(
            all_input_ids[prefix_offset:read_offset],
            skip_special_tokens=False,
            clean_up_tokenization_spaces=False,
        )
        new_text = self.decode(
            all_input_ids[prefix_offset:],
            skip_special_tokens=False,
            clean_up_tokenization_spaces=False,
        )

        if (
            len(new_text) > len(prefix_text)
            and not prefix_text.endswith("�")
            and not new_text.endswith("�")
        ):
            prefix_index = new_text.index(prefix_text)
            new_text = new_text[prefix_index + len(prefix_text) :]
            return new_text, read_offset, len(all_input_ids)
        else:
            return "", prefix_offset, read_offset

    def batch_decode(
        self,
        sequences,
        skip_special_tokens: bool = False,
        clean_up_tokenization_spaces: bool = True,
        **kwargs,
    ) -> List[str]:
        """
        Convert a list of lists of token ids into a list of strings by calling decode.

        Args:
            sequences (`Union[List[int], List[List[int]], np.ndarray, paddle.Tensor]`):
                List of tokenized input ids. Can be obtained using the `__call__` method.
            skip_special_tokens (`bool`, *optional*, defaults to `False`):
                Whether or not to remove special tokens in the decoding.
            clean_up_tokenization_spaces (`bool`, *optional*, defaults to `True`):
                Whether or not to clean up the tokenization spaces.
            kwargs (additional keyword arguments, *optional*):
                Will be passed to the underlying model specific decode method.

        Returns:
            `List[str]`: The list of decoded sentences.
        """
        return [
            self.decode(
                seq,
                skip_special_tokens=skip_special_tokens,
                clean_up_tokenization_spaces=clean_up_tokenization_spaces,
                **kwargs,
            )
            for seq in sequences
        ]

    def decode(
        self,
        token_ids,
        skip_special_tokens: bool = False,
        clean_up_tokenization_spaces: bool = True,
        **kwargs,
    ) -> str:
        """
        Converts a sequence of ids in a string, using the tokenizer and vocabulary with options to remove special
        tokens and clean up tokenization spaces.

        Similar to doing `self.convert_tokens_to_string(self.convert_ids_to_tokens(token_ids))`.

        Args:
            token_ids (`Union[int, List[int], np.ndarray, paddle.Tensor]`):
                List of tokenized input ids. Can be obtained using the `__call__` method.
            skip_special_tokens (`bool`, *optional*, defaults to `False`):
                Whether or not to remove special tokens in the decoding.
            clean_up_tokenization_spaces (`bool`, *optional*, defaults to `True`):
                Whether or not to clean up the tokenization spaces.
            kwargs (additional keyword arguments, *optional*):
                Will be passed to the underlying model specific decode method.

        Returns:
            `str`: The decoded sentence.
        """
        # Convert inputs to python lists
        token_ids = to_py_obj(token_ids)

        return self._decode(
            token_ids=token_ids,
            skip_special_tokens=skip_special_tokens,
            clean_up_tokenization_spaces=clean_up_tokenization_spaces,
            **kwargs,
        )

    def _decode(
        self,
        token_ids: Union[int, List[int]],
        skip_special_tokens: bool = False,
        clean_up_tokenization_spaces: bool = True,
        **kwargs,
    ) -> str:
        raise NotImplementedError

    def get_special_tokens_mask(
        self,
        token_ids_0: List[int],
        token_ids_1: Optional[List[int]] = None,
        already_has_special_tokens: bool = False,
    ) -> List[int]:
        """
        Retrieves sequence ids from a token list that has no special tokens added. This method is called when adding
        special tokens using the tokenizer `prepare_for_model` or `encode_plus` methods.

        Args:
            token_ids_0 (`List[int]`):
                List of ids of the first sequence.
            token_ids_1 (`List[int]`, *optional*):
                List of ids of the second sequence.
            already_has_special_tokens (`bool`, *optional*, defaults to `False`):
                Whether or not the token list is already formatted with special tokens for the model.

        Returns:
            A list of integers in the range [0, 1]: 1 for a special token, 0 for a sequence token.
        """
        assert already_has_special_tokens and token_ids_1 is None, (
            "You cannot use ``already_has_special_tokens=False`` with this tokenizer. "
            "Please use a slow (full python) tokenizer to activate this argument. "
            "Or set `return_special_tokens_mask=True` when calling the encoding method "
            "to get the special tokens mask in any tokenizer. "
        )

        all_special_ids = self.all_special_ids  # cache the property

        special_tokens_mask = [
            1 if token in all_special_ids else 0 for token in token_ids_0
        ]

        return special_tokens_mask

    @staticmethod
    def clean_up_tokenization(out_string: str) -> str:
        """
        Clean up a list of simple English tokenization artifacts like spaces before punctuations and abbreviated forms.

        Args:
            out_string (`str`): The text to clean up.

        Returns:
            `str`: The cleaned-up string.
        """
        out_string = (
            out_string.replace(" .", ".")
            .replace(" ?", "?")
            .replace(" !", "!")
            .replace(" ,", ",")
            .replace(" ' ", "'")
            .replace(" n't", "n't")
            .replace(" 'm", "'m")
            .replace(" 's", "'s")
            .replace(" 've", "'ve")
            .replace(" 're", "'re")
        )
        return out_string

    def _eventual_warn_about_too_long_sequence(
        self, ids: List[int], max_length: Optional[int], verbose: bool
    ):
        """
        Depending on the input and internal state we might trigger a warning about a sequence that is too long for its
        corresponding model

        Args:
            ids (`List[str]`): The ids produced by the tokenization
            max_length (`int`, *optional*): The max_length desired (does not trigger a warning if it is set)
            verbose (`bool`): Whether or not to print more information and warnings.

        """
        if max_length is None and len(ids) > self.model_max_length and verbose:
            if not self.deprecation_warnings.get(
                "sequence-length-is-longer-than-the-specified-maximum", False
            ):
                logging.warning(
                    "Token indices sequence length is longer than the specified maximum sequence length "
                    f"for this model ({len(ids)} > {self.model_max_length}). Running this sequence through the model "
                    "will result in indexing errors"
                )
            self.deprecation_warnings[
                "sequence-length-is-longer-than-the-specified-maximum"
            ] = True