Source code for paddlenlp.transformers.deberta_v2.tokenizer

# Copyright (c) 2023 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 os
import shutil
import unicodedata
from typing import Any, Dict, List, Optional, Tuple

import sentencepiece as sp

from .. import AddedToken, PretrainedTokenizer

__all__ = ["DebertaV2Tokenizer"]


PRETRAINED_VOCAB_FILES_MAP = {
    "sentencepiece_model_file": {
        "microsoft/deberta-v2-xlarge": "https://paddlenlp.bj.bcebos.com/models/community/microsoft/deberta-v2-xlarge/spm.model",
        "microsoft/deberta-v3-base": "https://paddlenlp.bj.bcebos.com/models/community/microsoft/deberta-v3-base/spm.model",
        "microsoft/deberta-v3-large": "https://paddlenlp.bj.bcebos.com/models/community/microsoft/deberta-v3-large/spm.model",
        "deepset/deberta-v3-large-squad2": "https://paddlenlp.bj.bcebos.com/models/community/deepset/deberta-v3-large-squad2/spm.model",
    }
}


PRETRAINED_POSITIONAL_EMBEDDINGS_SIZES = {
    "microsoft/deberta-v2-xlarge": 512,
    "microsoft/deberta-v3-base": 512,
    "microsoft/deberta-v3-large": 512,
    "deepset/deberta-v3-large-squad2": 512,
}

PRETRAINED_INIT_CONFIGURATION = {
    "microsoft/deberta-v2-xlarge": {"do_lower_case": False},
    "microsoft/deberta-v3-base": {"do_lower_case": False},
    "microsoft/deberta-v3-large": {"do_lower_case": False},
    "deepset/deberta-v3-large-squad2": {"do_lower_case": False},
}


[docs]class DebertaV2Tokenizer(PretrainedTokenizer): r""" Constructs a DeBERTa-v2 tokenizer. Based on [SentencePiece](https://github.com/google/sentencepiece). Args: vocab_file (`str`): [SentencePiece](https://github.com/google/sentencepiece) file (generally has a *.spm* extension) that contains the vocabulary necessary to instantiate a tokenizer. do_lower_case (`bool`, *optional*, defaults to `False`): Whether or not to lowercase the input when tokenizing. bos_token (`string`, *optional*, defaults to `"[CLS]"`): The beginning of sequence token that was used during pre-training. Can be used a sequence classifier token. When building a sequence using special tokens, this is not the token that is used for the beginning of sequence. The token used is the `cls_token`. eos_token (`string`, *optional*, defaults to `"[SEP]"`): The end of sequence token. When building a sequence using special tokens, this is not the token that is used for the end of sequence. The token used is the `sep_token`. unk_token (`str`, *optional*, defaults to `"[UNK]"`): The unknown token. A token that is not in the vocabulary cannot be converted to an ID and is set to be this token instead. sep_token (`str`, *optional*, defaults to `"[SEP]"`): The separator token, which is used when building a sequence from multiple sequences, e.g. two sequences for sequence classification or for a text and a question for question answering. It is also used as the last token of a sequence built with special tokens. pad_token (`str`, *optional*, defaults to `"[PAD]"`): The token used for padding, for example when batching sequences of different lengths. cls_token (`str`, *optional*, defaults to `"[CLS]"`): The classifier token which is used when doing sequence classification (classification of the whole sequence instead of per-token classification). It is the first token of the sequence when built with special tokens. mask_token (`str`, *optional*, defaults to `"[MASK]"`): The token used for masking values. This is the token used when training this model with masked language modeling. This is the token which the model will try to predict. sp_model_kwargs (`dict`, *optional*): Will be passed to the `SentencePieceProcessor.__init__()` method. The [Python wrapper for SentencePiece](https://github.com/google/sentencepiece/tree/master/python) can be used, among other things, to set: - `enable_sampling`: Enable subword regularization. - `nbest_size`: Sampling parameters for unigram. Invalid for BPE-Dropout. - `nbest_size = {0,1}`: No sampling is performed. - `nbest_size > 1`: samples from the nbest_size results. - `nbest_size < 0`: assuming that nbest_size is infinite and samples from the all hypothesis (lattice) using forward-filtering-and-backward-sampling algorithm. - `alpha`: Smoothing parameter for unigram sampling, and dropout probability of merge operations for BPE-dropout. """ resource_files_names = {"sentencepiece_model_file": "spm.model"} pretrained_resource_files_map = PRETRAINED_VOCAB_FILES_MAP pretrained_init_configuration = PRETRAINED_INIT_CONFIGURATION max_model_input_sizes = PRETRAINED_POSITIONAL_EMBEDDINGS_SIZES def __init__( self, sentencepiece_model_file, vocab_file=None, do_lower_case=False, split_by_punct=False, bos_token="[CLS]", eos_token="[SEP]", unk_token="[UNK]", sep_token="[SEP]", pad_token="[PAD]", cls_token="[CLS]", mask_token="[MASK]", sp_model_kwargs: Optional[Dict[str, Any]] = None, **kwargs, ) -> None: self.sp_model_kwargs = {} if sp_model_kwargs is None else sp_model_kwargs bos_token = AddedToken(bos_token, lstrip=False, rstrip=False) if isinstance(bos_token, str) else bos_token eos_token = AddedToken(eos_token, lstrip=False, rstrip=False) if isinstance(eos_token, str) else eos_token sep_token = AddedToken(sep_token, lstrip=False, rstrip=False) if isinstance(sep_token, str) else sep_token cls_token = AddedToken(cls_token, lstrip=False, rstrip=False) if isinstance(cls_token, str) else cls_token pad_token = AddedToken(pad_token, lstrip=False, rstrip=False) if isinstance(pad_token, str) else pad_token eos_token = AddedToken(eos_token, lstrip=False, rstrip=False) if isinstance(eos_token, str) else eos_token unk_token = AddedToken(unk_token, lstrip=False, rstrip=False) if isinstance(unk_token, str) else unk_token mask_token = AddedToken(mask_token, lstrip=False, rstrip=False) if isinstance(mask_token, str) else mask_token self._build_special_tokens_map_extended( bos_token=bos_token, eos_token=eos_token, sep_token=sep_token, cls_token=cls_token, pad_token=pad_token, mask_token=mask_token, unk_token=unk_token, ) self._sentencepiece_model_file = sentencepiece_model_file if vocab_file is None: self.vocab_file = sentencepiece_model_file else: self.vocab_file = vocab_file if not os.path.isfile(self.vocab_file): raise ValueError( f"Can't find a vocabulary file at path '{self.vocab_file}'. To load the vocabulary from a Google pretrained" " model use `tokenizer = AutoTokenizer.from_pretrained(PRETRAINED_MODEL_NAME)`" ) self.do_lower_case = do_lower_case self.split_by_punct = split_by_punct # breakpoint() self._tokenizer = SPMTokenizer( self.vocab_file, self.all_special_tokens, split_by_punct=split_by_punct, sp_model_kwargs=self.sp_model_kwargs, ) @property def vocab_size(self): return len(self.vocab) @property def vocab(self): return self._tokenizer.vocab
[docs] def get_vocab(self): vocab = self.vocab.copy() vocab.update(self.get_added_vocab()) return vocab
def _tokenize(self, text: str) -> List[str]: """Take as input a string and return a list of strings (tokens) for words/sub-words""" if self.do_lower_case: text = text.lower() return self._tokenizer.tokenize(text) def _convert_token_to_id(self, token): """Converts a token (str) in an id using the vocab.""" return self._tokenizer.spm.PieceToId(token) def _convert_id_to_token(self, index): """Converts an index (integer) in a token (str) using the vocab.""" return self._tokenizer.spm.IdToPiece(index) if index < self.vocab_size else self.unk_token
[docs] def convert_tokens_to_string(self, tokens): """Converts a sequence of tokens (string) in a single string.""" return self._tokenizer.decode(tokens)
[docs] def build_inputs_with_special_tokens(self, token_ids_0, token_ids_1=None): """ Build model inputs from a sequence or a pair of sequence for sequence classification tasks by concatenating and adding special tokens. A DeBERTa sequence has the following format: - single sequence: [CLS] X [SEP] - pair of sequences: [CLS] A [SEP] B [SEP] Args: token_ids_0 (`List[int]`): List of IDs to which the special tokens will be added. token_ids_1 (`List[int]`, *optional*): Optional second list of IDs for sequence pairs. Returns: `List[int]`: List of [input IDs](../glossary#input-ids) with the appropriate special tokens. """ if token_ids_1 is None: return [self.cls_token_id] + token_ids_0 + [self.sep_token_id] cls = [self.cls_token_id] sep = [self.sep_token_id] return cls + token_ids_0 + sep + token_ids_1 + sep
[docs] def get_special_tokens_mask(self, token_ids_0, token_ids_1=None, already_has_special_tokens=False): """ 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. token_ids_1 (`List[int]`, *optional*): Optional second list of IDs for sequence pairs. 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: `List[int]`: A list of integers in the range [0, 1]: 1 for a special token, 0 for a sequence token. """ if already_has_special_tokens: return super().get_special_tokens_mask( token_ids_0=token_ids_0, token_ids_1=token_ids_1, already_has_special_tokens=True ) if token_ids_1 is not None: return [1] + ([0] * len(token_ids_0)) + [1] + ([0] * len(token_ids_1)) + [1] return [1] + ([0] * len(token_ids_0)) + [1]
[docs] def create_token_type_ids_from_sequences(self, token_ids_0, token_ids_1=None): """ Create a mask from the two sequences passed to be used in a sequence-pair classification task. A DeBERTa sequence pair mask has the following format: ``` 0 0 0 0 0 0 0 0 0 0 0 1 1 1 1 1 1 1 1 1 | first sequence | second sequence | ``` If `token_ids_1` is `None`, this method only returns the first portion of the mask (0s). Args: token_ids_0 (`List[int]`): List of IDs. token_ids_1 (`List[int]`, *optional*): Optional second list of IDs for sequence pairs. Returns: `List[int]`: List of [token type IDs](../glossary#token-type-ids) according to the given sequence(s). """ sep = [self.sep_token_id] cls = [self.cls_token_id] if token_ids_1 is None: return len(cls + token_ids_0 + sep) * [0] return len(cls + token_ids_0 + sep) * [0] + len(token_ids_1 + sep) * [1]
[docs] def prepare_for_tokenization(self, text, is_split_into_words=False, **kwargs): add_prefix_space = kwargs.pop("add_prefix_space", False) if is_split_into_words or add_prefix_space: text = " " + text return (text, kwargs)
[docs] def save_vocabulary(self, save_directory: str, filename_prefix: Optional[str] = None) -> Tuple[str]: return self._tokenizer.save_pretrained(save_directory, filename_prefix=filename_prefix)
[docs] 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. A BERT offset_mapping has the following format: - single sequence: ``(0,0) X (0,0)`` - pair of sequences: ``(0,0) A (0,0) B (0,0)`` Args: offset_mapping_ids_0 (List[tuple]): List of wordpiece offsets to which the special tokens will be added. offset_mapping_ids_1 (List[tuple], optional): Optional second list of wordpiece offsets for offset mapping pairs. Defaults to None. Returns: List[tuple]: A list of wordpiece offsets with the appropriate offsets of special tokens. """ if offset_mapping_1 is None: return [(0, 0)] + offset_mapping_0 + [(0, 0)] return [(0, 0)] + offset_mapping_0 + [(0, 0)] + offset_mapping_1 + [(0, 0)]
[docs] def save_resources(self, save_directory): """ Saves `SentencePiece <https://github.com/google/sentencepiece>`__ file (ends with '.spm') under `save_directory`. Args: save_directory (str): Directory to save files into. """ for name, file_name in self.resource_files_names.items(): source_path = getattr(self, "_%s" % name) save_path = os.path.join(save_directory, file_name) if os.path.abspath(source_path) != os.path.abspath(save_path): shutil.copyfile(source_path, save_path)
class SPMTokenizer: r""" Constructs a tokenizer based on [SentencePiece](https://github.com/google/sentencepiece). Args: vocab_file (`str`): [SentencePiece](https://github.com/google/sentencepiece) file (generally has a *.spm* extension) that contains the vocabulary necessary to instantiate a tokenizer. sp_model_kwargs (`dict`, *optional*): Will be passed to the `SentencePieceProcessor.__init__()` method. The [Python wrapper for SentencePiece](https://github.com/google/sentencepiece/tree/master/python) can be used, among other things, to set: - `enable_sampling`: Enable subword regularization. - `nbest_size`: Sampling parameters for unigram. Invalid for BPE-Dropout. - `nbest_size = {0,1}`: No sampling is performed. - `nbest_size > 1`: samples from the nbest_size results. - `nbest_size < 0`: assuming that nbest_size is infinite and samples from the all hypothesis (lattice) using forward-filtering-and-backward-sampling algorithm. - `alpha`: Smoothing parameter for unigram sampling, and dropout probability of merge operations for BPE-dropout. """ def __init__( self, vocab_file, special_tokens, split_by_punct=False, sp_model_kwargs: Optional[Dict[str, Any]] = None ): self.split_by_punct = split_by_punct self.vocab_file = vocab_file self.sp_model_kwargs = {} if sp_model_kwargs is None else sp_model_kwargs spm = sp.SentencePieceProcessor(**self.sp_model_kwargs) if not os.path.exists(vocab_file): raise FileNotFoundError(f"{vocab_file} does not exist!") spm.load(vocab_file) bpe_vocab_size = spm.GetPieceSize() # Token map # <unk> 0+1 # <s> 1+1 # </s> 2+1 self.vocab = {spm.IdToPiece(i): i for i in range(bpe_vocab_size)} self.ids_to_tokens = [spm.IdToPiece(i) for i in range(bpe_vocab_size)] # self.vocab['[PAD]'] = 0 # self.vocab['[CLS]'] = 1 # self.vocab['[SEP]'] = 2 # self.vocab['[UNK]'] = 3 self.spm = spm self.special_tokens = special_tokens def __getstate__(self): state = self.__dict__.copy() state["spm"] = None return state def __setstate__(self, d): self.__dict__ = d # for backward compatibility if not hasattr(self, "sp_model_kwargs"): self.sp_model_kwargs = {} self.spm = sp.SentencePieceProcessor(**self.sp_model_kwargs) self.spm.Load(self.vocab_file) def tokenize(self, text): return self._encode_as_pieces(text) def convert_ids_to_tokens(self, ids): tokens = [] for i in ids: tokens.append(self.ids_to_tokens[i]) return tokens def decode(self, tokens, start=-1, end=-1, raw_text=None): if raw_text is None: current_sub_tokens = [] out_string = "" prev_is_special = False for token in tokens: # make sure that special tokens are not decoded using sentencepiece model if token in self.special_tokens: if not prev_is_special: out_string += " " out_string += self.spm.decode_pieces(current_sub_tokens) + token prev_is_special = True current_sub_tokens = [] else: current_sub_tokens.append(token) prev_is_special = False out_string += self.spm.decode_pieces(current_sub_tokens) return out_string.strip() else: words = self.split_to_words(raw_text) word_tokens = [self.tokenize(w) for w in words] token2words = [0] * len(tokens) tid = 0 for i, w in enumerate(word_tokens): for k, t in enumerate(w): token2words[tid] = i tid += 1 word_start = token2words[start] word_end = token2words[end] if end < len(tokens) else len(words) text = "".join(words[word_start:word_end]) return text def add_special_token(self, token): if token not in self.special_tokens: self.special_tokens.append(token) if token not in self.vocab: self.vocab[token] = len(self.vocab) - 1 self.ids_to_tokens.append(token) return self.id(token) def part_of_whole_word(self, token, is_bos=False): if is_bos: return True if ( len(token) == 1 and (_is_whitespace(list(token)[0]) or _is_control(list(token)[0]) or _is_punctuation(list(token)[0])) ) or token in self.special_tokens: return False word_start = b"\xe2\x96\x81".decode("utf-8") return not token.startswith(word_start) def pad(self): return "[PAD]" def bos(self): return "[CLS]" def eos(self): return "[SEP]" def unk(self): return "[UNK]" def mask(self): return "[MASK]" def sym(self, id): return self.ids_to_tokens[id] def id(self, sym): return self.vocab[sym] if sym in self.vocab else 1 def _encode_as_pieces(self, text): text = convert_to_unicode(text) if self.split_by_punct: words = self._run_split_on_punc(text) pieces = [self.spm.encode(w, out_type=str) for w in words] return [p for w in pieces for p in w] else: return self.spm.encode(text, out_type=str) def split_to_words(self, text): pieces = self._encode_as_pieces(text) word_start = b"\xe2\x96\x81".decode("utf-8") words = [] offset = 0 prev_end = 0 for i, p in enumerate(pieces): if p.startswith(word_start): if offset > prev_end: words.append(text[prev_end:offset]) prev_end = offset w = p.replace(word_start, "") else: w = p try: s = text.index(w, offset) pn = "" k = i + 1 while k < len(pieces): pn = pieces[k].replace(word_start, "") if len(pn) > 0: break k += 1 if len(pn) > 0 and pn in text[offset:s]: offset = offset + 1 else: offset = s + len(w) except Exception: offset = offset + 1 if prev_end < offset: words.append(text[prev_end:offset]) return words def _run_strip_accents(self, text): """Strips accents from a piece of text.""" text = unicodedata.normalize("NFD", text) output = [] for char in text: cat = unicodedata.category(char) if cat == "Mn": continue output.append(char) return "".join(output) def _run_split_on_punc(self, text): """Splits punctuation on a piece of text.""" chars = list(text) i = 0 start_new_word = True output = [] while i < len(chars): char = chars[i] if _is_punctuation(char): output.append([char]) start_new_word = True else: if start_new_word: output.append([]) start_new_word = False output[-1].append(char) i += 1 return ["".join(x) for x in output] def _is_whitespace(char): """Checks whether `chars` is a whitespace character.""" # \t, \n, and \r are technically control characters but we treat them # as whitespace since they are generally considered as such. if char == " " or char == "\t" or char == "\n" or char == "\r": return True cat = unicodedata.category(char) if cat == "Zs": return True return False def _is_control(char): """Checks whether `chars` is a control character.""" # These are technically control characters but we count them as whitespace # characters. if char == "\t" or char == "\n" or char == "\r": return False cat = unicodedata.category(char) if cat.startswith("C"): return True return False def _is_punctuation(char): """Checks whether `chars` is a punctuation character.""" cp = ord(char) # We treat all non-letter/number ASCII as punctuation. # Characters such as "^", "$", and "`" are not in the Unicode # Punctuation class but we treat them as punctuation anyways, for # consistency. if (cp >= 33 and cp <= 47) or (cp >= 58 and cp <= 64) or (cp >= 91 and cp <= 96) or (cp >= 123 and cp <= 126): return True cat = unicodedata.category(char) if cat.startswith("P"): return True return False def convert_to_unicode(text): """Converts `text` to Unicode (if it's not already), assuming utf-8 input.""" if isinstance(text, str): return text elif isinstance(text, bytes): return text.decode("utf-8", "ignore") else: raise ValueError(f"Unsupported string type: {type(text)}")