# coding=utf-8
# Copyright 2021 The HuggingFace Inc. team.
#
# 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
""" Tokenization classes for LayoutXLM model."""
import os
from shutil import copyfile
from typing import Any, Dict, List, Optional, Tuple, Union
import sentencepiece as spm
from transformers.models.layoutlmv2.tokenization_layoutlmv2 import LAYOUTLMV2_ENCODE_PLUS_ADDITIONAL_KWARGS_DOCSTRING
from ...file_utils import PaddingStrategy, TensorType, add_end_docstrings
from ...tokenization_utils import AddedToken, PreTrainedTokenizer
from ...tokenization_utils_base import (
ENCODE_KWARGS_DOCSTRING,
BatchEncoding,
EncodedInput,
PreTokenizedInput,
TextInput,
TextInputPair,
TruncationStrategy,
)
from ...utils import logging
from ..xlm_roberta.tokenization_xlm_roberta import (
PRETRAINED_POSITIONAL_EMBEDDINGS_SIZES,
PRETRAINED_VOCAB_FILES_MAP,
SPIECE_UNDERLINE,
VOCAB_FILES_NAMES,
)
logger = logging.get_logger(__name__)
class LayoutXLMTokenizer(PreTrainedTokenizer):
"""
Adapted from [`RobertaTokenizer`] and [`XLNetTokenizer`]. Based on
[SentencePiece](https://github.com/google/sentencepiece).
This tokenizer inherits from [`PreTrainedTokenizer`] which contains most of the main methods.
Users should refer to this superclass for more information regarding those methods.
Args:
vocab_file (`str`):
Path to the vocabulary file.
bos_token (`str`, *optional*, defaults to `""`):
The beginning of sequence token that was used during pretraining. 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 (`str`, *optional*, defaults to `""`):
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`.
sep_token (`str`, *optional*, defaults to `""`):
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.
cls_token (`str`, *optional*, defaults to `""`):
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.
unk_token (`str`, *optional*, defaults to `""`):
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.
pad_token (`str`, *optional*, defaults to `""`):
The token used for padding, for example when batching sequences of different lengths.
mask_token (`str`, *optional*, defaults to `""`):
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.
cls_token_box (`List[int]`, *optional*, defaults to `[0, 0, 0, 0]`):
The bounding box to use for the special [CLS] token.
sep_token_box (`List[int]`, *optional*, defaults to `[1000, 1000, 1000, 1000]`):
The bounding box to use for the special [SEP] token.
pad_token_box (`List[int]`, *optional*, defaults to `[0, 0, 0, 0]`):
The bounding box to use for the special [PAD] token.
pad_token_label (`int`, *optional*, defaults to -100):
The label to use for padding tokens. Defaults to -100, which is the `ignore_index` of PyTorch's
CrossEntropyLoss.
only_label_first_subword (`bool`, *optional*, defaults to `True`):
Whether or not to only label the first subword, in case word labels are provided.
additional_special_tokens (`List[str]`, *optional*, defaults to `["NOTUSED", "NOTUSED"]`):
Additional special tokens used by the 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.
Attributes:
sp_model (`SentencePieceProcessor`):
The *SentencePiece* processor that is used for every conversion (string, tokens and IDs).
"""
vocab_files_names = VOCAB_FILES_NAMES
pretrained_vocab_files_map = PRETRAINED_VOCAB_FILES_MAP
max_model_input_sizes = PRETRAINED_POSITIONAL_EMBEDDINGS_SIZES
model_input_names = ["input_ids", "attention_mask"]
def __init__(
self,
vocab_file,
bos_token="",
eos_token="",
sep_token="",
cls_token="",
unk_token="",
pad_token="",
mask_token="",
cls_token_box=[0, 0, 0, 0],
sep_token_box=[1000, 1000, 1000, 1000],
pad_token_box=[0, 0, 0, 0],
pad_token_label=-100,
only_label_first_subword=True,
sp_model_kwargs: Optional[Dict[str, Any]] = None,
**kwargs
) -> None:
# Mask token behave like a normal word, i.e. include the space before it
mask_token = AddedToken(mask_token, lstrip=True, rstrip=False) if isinstance(mask_token, str) else mask_token
self.sp_model_kwargs = {} if sp_model_kwargs is None else sp_model_kwargs
super().__init__(
bos_token=bos_token,
eos_token=eos_token,
unk_token=unk_token,
sep_token=sep_token,
cls_token=cls_token,
pad_token=pad_token,
mask_token=mask_token,
cls_token_box=cls_token_box,
sep_token_box=sep_token_box,
pad_token_box=pad_token_box,
pad_token_label=pad_token_label,
only_label_first_subword=only_label_first_subword,
sp_model_kwargs=self.sp_model_kwargs,
**kwargs,
)
self.sp_model = spm.SentencePieceProcessor(**self.sp_model_kwargs)
self.sp_model.Load(str(vocab_file))
self.vocab_file = vocab_file
# Original fairseq vocab and spm vocab must be "aligned":
# Vocab | 0 | 1 | 2 | 3 | 4 | 5 | 6 | 7 | 8 | 9
# -------- | ------- | ------- | ------ | ------- | --- | --- | --- | ----- | ----- | ----
# fairseq | '' | '' | '' | '' | ',' | '.' | '▁' | 's' | '▁de' | '-'
# spm | '' | '' | '' | ',' | '.' | '▁' | 's' | '▁de' | '-' | '▁a'
# Mimic fairseq token-to-id alignment for the first 4 token
self.fairseq_tokens_to_ids = {"": 0, "": 1, "": 2, "": 3}
# The first "real" token "," has position 4 in the original fairseq vocab and position 3 in the spm vocab
self.fairseq_offset = 1
self.fairseq_tokens_to_ids[""] = len(self.sp_model) + self.fairseq_offset
self.fairseq_ids_to_tokens = {v: k for k, v in self.fairseq_tokens_to_ids.items()}
# additional properties
self.cls_token_box = cls_token_box
self.sep_token_box = sep_token_box
self.pad_token_box = pad_token_box
self.pad_token_label = pad_token_label
self.only_label_first_subword = only_label_first_subword
def __getstate__(self):
state = self.__dict__.copy()
state["sp_model"] = None
state["sp_model_proto"] = self.sp_model.serialized_model_proto()
return state
def __setstate__(self, d):
self.__dict__ = d
# for backward compatibility
if not hasattr(self, "sp_model_kwargs"):
self.sp_model_kwargs = {}
self.sp_model = spm.SentencePieceProcessor(**self.sp_model_kwargs)
self.sp_model.LoadFromSerializedProto(self.sp_model_proto)
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. An XLM-RoBERTa sequence has the following format:
- single sequence: ` X `
- pair of sequences: ` A B `
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 + sep + token_ids_1 + sep
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]:
"""
Retrieve 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` method.
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 None:
return [1] + ([0] * len(token_ids_0)) + [1]
return [1] + ([0] * len(token_ids_0)) + [1, 1] + ([0] * len(token_ids_1)) + [1]
def create_token_type_ids_from_sequences(
self, token_ids_0: List[int], token_ids_1: Optional[List[int]] = None
) -> List[int]:
"""
Create a mask from the two sequences passed to be used in a sequence-pair classification task. XLM-RoBERTa does
not make use of token type ids, therefore a list of zeros is returned.
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 zeros.
"""
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 + sep + token_ids_1 + sep) * [0]
@property
def vocab_size(self):
return len(self.sp_model) + self.fairseq_offset + 1 # Add the token
def get_vocab(self):
vocab = {self.convert_ids_to_tokens(i): i for i in range(self.vocab_size)}
vocab.update(self.added_tokens_encoder)
return vocab
def _tokenize(self, text: str) -> List[str]:
return self.sp_model.encode(text, out_type=str)
def _convert_token_to_id(self, token):
"""Converts a token (str) in an id using the vocab."""
if token in self.fairseq_tokens_to_ids:
return self.fairseq_tokens_to_ids[token]
spm_id = self.sp_model.PieceToId(token)
# Need to return unknown token if the SP model returned 0
return spm_id + self.fairseq_offset if spm_id else self.unk_token_id
def _convert_id_to_token(self, index):
"""Converts an index (integer) in a token (str) using the vocab."""
if index in self.fairseq_ids_to_tokens:
return self.fairseq_ids_to_tokens[index]
return self.sp_model.IdToPiece(index - self.fairseq_offset)
def convert_tokens_to_string(self, tokens):
"""Converts a sequence of tokens (strings for sub-words) in a single string."""
out_string = "".join(tokens).replace(SPIECE_UNDERLINE, " ").strip()
return out_string
def save_vocabulary(self, save_directory: str, filename_prefix: Optional[str] = None) -> Tuple[str]:
if not os.path.isdir(save_directory):
logger.error(f"Vocabulary path ({save_directory}) should be a directory")
return
out_vocab_file = os.path.join(
save_directory, (filename_prefix + "-" if filename_prefix else "") + VOCAB_FILES_NAMES["vocab_file"]
)
if os.path.abspath(self.vocab_file) != os.path.abspath(out_vocab_file):
copyfile(self.vocab_file, out_vocab_file)
return (out_vocab_file,)
@add_end_docstrings(ENCODE_KWARGS_DOCSTRING, LAYOUTLMV2_ENCODE_PLUS_ADDITIONAL_KWARGS_DOCSTRING)
def __call__(
self,
text: Union[TextInput, PreTokenizedInput, List[TextInput], List[PreTokenizedInput]],
text_pair: Optional[Union[PreTokenizedInput, List[PreTokenizedInput]]] = None,
boxes: Union[List[List[int]], List[List[List[int]]]] = None,
word_labels: Optional[Union[List[int], List[List[int]]]] = None,
add_special_tokens: bool = True,
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,
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:
"""
Main method to tokenize and prepare for the model one or several sequence(s) or one or several pair(s) of
sequences with word-level normalized bounding boxes and optional labels.
Args:
text (`str`, `List[str]`, `List[List[str]]`):
The sequence or batch of sequences to be encoded. Each sequence can be a string, a list of strings
(words of a single example or questions of a batch of examples) or a list of list of strings (batch of
words).
text_pair (`List[str]`, `List[List[str]]`):
The sequence or batch of sequences to be encoded. Each sequence should be a list of strings
(pretokenized string).
boxes (`List[List[int]]`, `List[List[List[int]]]`):
Word-level bounding boxes. Each bounding box should be normalized to be on a 0-1000 scale.
word_labels (`List[int]`, `List[List[int]]`, *optional*):
Word-level integer labels (for token classification tasks such as FUNSD, CORD).
"""
# 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 text_pair is not None:
# in case text + text_pair are provided, text = questions, text_pair = words
if not _is_valid_text_input(text):
raise ValueError("text input must of type `str` (single example) or `List[str]` (batch of examples). ")
if not isinstance(text_pair, (list, tuple)):
raise ValueError(
"words must of type `List[str]` (single pretokenized example), "
"or `List[List[str]]` (batch of pretokenized examples)."
)
else:
# in case only text is provided => must be words
if not isinstance(text, (list, tuple)):
raise ValueError(
"Words must of type `List[str]` (single pretokenized example), "
"or `List[List[str]]` (batch of pretokenized examples)."
)
if text_pair is not None:
is_batched = isinstance(text, (list, tuple))
else:
is_batched = isinstance(text, (list, tuple)) and text and isinstance(text[0], (list, tuple))
words = text if text_pair is None else text_pair
if boxes is None:
raise ValueError("You must provide corresponding bounding boxes")
if is_batched:
if len(words) != len(boxes):
raise ValueError("You must provide words and boxes for an equal amount of examples")
for words_example, boxes_example in zip(words, boxes):
if len(words_example) != len(boxes_example):
raise ValueError("You must provide as many words as there are bounding boxes")
else:
if len(words) != len(boxes):
raise ValueError("You must provide as many words as there are bounding boxes")
if is_batched:
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
is_pair = bool(text_pair is not None)
return self.batch_encode_plus(
batch_text_or_text_pairs=batch_text_or_text_pairs,
is_pair=is_pair,
boxes=boxes,
word_labels=word_labels,
add_special_tokens=add_special_tokens,
padding=padding,
truncation=truncation,
max_length=max_length,
stride=stride,
pad_to_multiple_of=pad_to_multiple_of,
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,
)
else:
return self.encode_plus(
text=text,
text_pair=text_pair,
boxes=boxes,
word_labels=word_labels,
add_special_tokens=add_special_tokens,
padding=padding,
truncation=truncation,
max_length=max_length,
stride=stride,
pad_to_multiple_of=pad_to_multiple_of,
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 _batch_encode_plus(
self,
batch_text_or_text_pairs: Union[
List[TextInput],
List[TextInputPair],
List[PreTokenizedInput],
],
is_pair: bool = None,
boxes: Optional[List[List[List[int]]]] = None,
word_labels: Optional[List[List[int]]] = 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,
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:
if return_offsets_mapping:
raise NotImplementedError(
"return_offset_mapping is not available when using Python tokenizers. "
"To use this feature, change your tokenizer to one deriving from "
"transformers.PreTrainedTokenizerFast."
)
batch_outputs = self._batch_prepare_for_model(
batch_text_or_text_pairs=batch_text_or_text_pairs,
is_pair=is_pair,
boxes=boxes,
word_labels=word_labels,
add_special_tokens=add_special_tokens,
padding_strategy=padding_strategy,
truncation_strategy=truncation_strategy,
max_length=max_length,
stride=stride,
pad_to_multiple_of=pad_to_multiple_of,
return_attention_mask=return_attention_mask,
return_token_type_ids=return_token_type_ids,
return_overflowing_tokens=return_overflowing_tokens,
return_special_tokens_mask=return_special_tokens_mask,
return_length=return_length,
return_tensors=return_tensors,
verbose=verbose,
)
return BatchEncoding(batch_outputs)
@add_end_docstrings(ENCODE_KWARGS_DOCSTRING, LAYOUTLMV2_ENCODE_PLUS_ADDITIONAL_KWARGS_DOCSTRING)
def _batch_prepare_for_model(
self,
batch_text_or_text_pairs,
is_pair: bool = None,
boxes: Optional[List[List[int]]] = None,
word_labels: Optional[List[List[int]]] = 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,
pad_to_multiple_of: Optional[int] = None,
return_tensors: Optional[str] = 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_length: bool = False,
verbose: bool = True,
) -> BatchEncoding:
"""
Prepares a sequence of input id, or a pair of sequences of inputs ids so that it can be used by the model. It
adds special tokens, truncates sequences if overflowing while taking into account the special tokens and
manages a moving window (with user defined stride) for overflowing tokens
Args:
batch_ids_pairs: list of tokenized input ids or input ids pairs
"""
batch_outputs = {}
for idx, example in enumerate(zip(batch_text_or_text_pairs, boxes)):
batch_text_or_text_pair, boxes_example = example
outputs = self.prepare_for_model(
batch_text_or_text_pair[0] if is_pair else batch_text_or_text_pair,
batch_text_or_text_pair[1] if is_pair else None,
boxes_example,
word_labels=word_labels[idx] if word_labels is not None else None,
add_special_tokens=add_special_tokens,
padding=PaddingStrategy.DO_NOT_PAD.value, # we pad in batch afterward
truncation=truncation_strategy.value,
max_length=max_length,
stride=stride,
pad_to_multiple_of=None, # we pad in batch afterward
return_attention_mask=False, # we pad in batch afterward
return_token_type_ids=return_token_type_ids,
return_overflowing_tokens=return_overflowing_tokens,
return_special_tokens_mask=return_special_tokens_mask,
return_length=return_length,
return_tensors=None, # We convert the whole batch to tensors at the end
prepend_batch_axis=False,
verbose=verbose,
)
for key, value in outputs.items():
if key not in batch_outputs:
batch_outputs[key] = []
batch_outputs[key].append(value)
batch_outputs = self.pad(
batch_outputs,
padding=padding_strategy.value,
max_length=max_length,
pad_to_multiple_of=pad_to_multiple_of,
return_attention_mask=return_attention_mask,
)
batch_outputs = BatchEncoding(batch_outputs, tensor_type=return_tensors)
return batch_outputs
def _encode_plus(
self,
text: Union[TextInput, PreTokenizedInput],
text_pair: Optional[PreTokenizedInput] = None,
boxes: Optional[List[List[int]]] = None,
word_labels: Optional[List[int]] = 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,
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:
if return_offsets_mapping:
raise NotImplementedError(
"return_offset_mapping is not available when using Python tokenizers. "
"To use this feature, change your tokenizer to one deriving from "
"transformers.PreTrainedTokenizerFast. "
"More information on available tokenizers at "
"https://github.com/huggingface/transformers/pull/2674"
)
return self.prepare_for_model(
text=text,
text_pair=text_pair,
boxes=boxes,
word_labels=word_labels,
add_special_tokens=add_special_tokens,
padding=padding_strategy.value,
truncation=truncation_strategy.value,
max_length=max_length,
stride=stride,
pad_to_multiple_of=pad_to_multiple_of,
return_tensors=return_tensors,
prepend_batch_axis=True,
return_attention_mask=return_attention_mask,
return_token_type_ids=return_token_type_ids,
return_overflowing_tokens=return_overflowing_tokens,
return_special_tokens_mask=return_special_tokens_mask,
return_length=return_length,
verbose=verbose,
)
@add_end_docstrings(ENCODE_KWARGS_DOCSTRING, LAYOUTLMV2_ENCODE_PLUS_ADDITIONAL_KWARGS_DOCSTRING)
def prepare_for_model(
self,
text: Union[TextInput, PreTokenizedInput],
text_pair: Optional[PreTokenizedInput] = None,
boxes: Optional[List[List[int]]] = None,
word_labels: Optional[List[int]] = None,
add_special_tokens: bool = True,
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,
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,
prepend_batch_axis: bool = False,
**kwargs
) -> BatchEncoding:
"""
Prepares a sequence or a pair of sequences so that it can be used by the model. It adds special tokens,
truncates sequences if overflowing while taking into account the special tokens and manages a moving window
(with user defined stride) for overflowing tokens.
Word-level `boxes` are turned into token-level `bbox`. If provided, word-level `word_labels` are
turned into token-level `labels`. The word label is used for the first token of the word, while remaining
tokens are labeled with -100, such that they will be ignored by the loss function.
Args:
text (`str`, `List[str]`, `List[List[str]]`):
The first sequence to be encoded. This can be a string, a list of strings or a list of list of strings.
text_pair (`List[str]` or `List[int]`, *optional*):
Optional second sequence to be encoded. This can be a list of strings (words of a single example) or a
list of list of strings (words of a batch of examples).
"""
# 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,
)
tokens = []
pair_tokens = []
token_boxes = []
pair_token_boxes = []
labels = []
if text_pair is None:
if word_labels is None:
# CASE 1: document image classification (training + inference) + CASE 2: token classification (inference)
for word, box in zip(text, boxes):
if len(word) < 1: # skip empty words
continue
word_tokens = self.tokenize(word)
tokens.extend(word_tokens)
token_boxes.extend([box] * len(word_tokens))
else:
# CASE 2: token classification (training)
for word, box, label in zip(text, boxes, word_labels):
if len(word) < 1: # skip empty words
continue
word_tokens = self.tokenize(word)
tokens.extend(word_tokens)
token_boxes.extend([box] * len(word_tokens))
if self.only_label_first_subword:
# Use the real label id for the first token of the word, and padding ids for the remaining tokens
labels.extend([label] + [self.pad_token_label] * (len(word_tokens) - 1))
else:
labels.extend([label] * len(word_tokens))
else:
# CASE 3: document visual question answering (inference)
# text = question
# text_pair = words
tokens = self.tokenize(text)
token_boxes = [self.pad_token_box for _ in range(len(tokens))] + [self.sep_token_box]
for word, box in zip(text_pair, boxes):
if len(word) < 1: # skip empty words
continue
word_tokens = self.tokenize(word)
pair_tokens.extend(word_tokens)
pair_token_boxes.extend([box] * len(word_tokens))
# Create ids + pair_ids
ids = self.convert_tokens_to_ids(tokens)
pair_ids = self.convert_tokens_to_ids(pair_tokens) if pair_tokens else None
# Compute the total size of the returned encodings
pair = bool(pair_ids is not None)
len_ids = len(ids)
len_pair_ids = len(pair_ids) if pair else 0
total_len = len_ids + len_pair_ids + (self.num_special_tokens_to_add(pair=pair) if add_special_tokens else 0)
# Truncation: Handle max sequence length
overflowing_tokens = []
overflowing_token_boxes = []
overflowing_labels = []
if truncation_strategy != TruncationStrategy.DO_NOT_TRUNCATE and max_length and total_len > max_length:
(
ids,
token_boxes,
pair_ids,
pair_token_boxes,
labels,
overflowing_tokens,
overflowing_token_boxes,
overflowing_labels,
) = self.truncate_sequences(
ids,
token_boxes,
pair_ids=pair_ids,
pair_token_boxes=pair_token_boxes,
labels=labels,
num_tokens_to_remove=total_len - max_length,
truncation_strategy=truncation_strategy,
stride=stride,
)
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."
)
# 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
encoded_inputs = {}
if return_overflowing_tokens:
encoded_inputs["overflowing_tokens"] = overflowing_tokens
encoded_inputs["overflowing_token_boxes"] = overflowing_token_boxes
encoded_inputs["overflowing_labels"] = overflowing_labels
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)
token_boxes = [self.cls_token_box] + token_boxes + [self.sep_token_box]
if pair_token_boxes:
pair_token_boxes = pair_token_boxes + [self.sep_token_box]
if labels:
labels = [self.pad_token_label] + labels + [self.pad_token_label]
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
encoded_inputs["bbox"] = token_boxes + pair_token_boxes
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 labels:
encoded_inputs["labels"] = labels
# Check lengths
self._eventual_warn_about_too_long_sequence(encoded_inputs["input_ids"], max_length, verbose)
# Padding
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,
return_attention_mask=return_attention_mask,
)
if return_length:
encoded_inputs["length"] = len(encoded_inputs["input_ids"])
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],
token_boxes: List[List[int]],
pair_ids: Optional[List[int]] = None,
pair_token_boxes: Optional[List[List[int]]] = None,
labels: 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.
token_boxes (`List[List[int]]`):
Bounding boxes of the first sequence.
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.
pair_token_boxes (`List[List[int]]`, *optional*):
Bounding boxes of the second sequence.
labels (`List[int]`, *optional*):
Labels of the first sequence (for token classification tasks).
num_tokens_to_remove (`int`, *optional*, defaults to 0):
Number of tokens to remove using the truncation strategy.
truncation_strategy (`str` or [`~tokenization_utils_base.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.
"""
if num_tokens_to_remove <= 0:
return ids, token_boxes, pair_ids, pair_token_boxes, labels, [], [], []
if not isinstance(truncation_strategy, TruncationStrategy):
truncation_strategy = TruncationStrategy(truncation_strategy)
overflowing_tokens = []
overflowing_token_boxes = []
overflowing_labels = []
if truncation_strategy == TruncationStrategy.LONGEST_FIRST:
for _ in range(num_tokens_to_remove):
if pair_ids is None or len(ids) > len(pair_ids):
if not overflowing_tokens:
window_len = min(len(ids), stride + 1)
else:
window_len = 1
overflowing_tokens.extend(ids[-window_len:])
overflowing_token_boxes.extend(token_boxes[-window_len:])
overflowing_labels.extend(labels[-window_len:])
ids = ids[:-1]
token_boxes = token_boxes[:-1]
labels = labels[:-1]
else:
if not overflowing_tokens:
window_len = min(len(pair_ids), stride + 1)
else:
window_len = 1
overflowing_tokens.extend(pair_ids[-window_len:])
overflowing_token_boxes.extend(pair_token_boxes[-window_len:])
pair_ids = pair_ids[:-1]
pair_token_boxes = pair_token_boxes[:-1]
elif truncation_strategy == TruncationStrategy.ONLY_FIRST:
if len(ids) > num_tokens_to_remove:
window_len = min(len(ids), stride + num_tokens_to_remove)
overflowing_tokens = ids[-window_len:]
overflowing_token_boxes = token_boxes[-window_len:]
overflowing_labels = labels[-window_len:]
ids = ids[:-num_tokens_to_remove]
token_boxes = token_boxes[:-num_tokens_to_remove]
labels = labels[:-num_tokens_to_remove]
else:
logger.error(
f"We need to remove {num_tokens_to_remove} to truncate the input "
f"but the first sequence has a length {len(ids)}. "
f"Please select another truncation strategy than {truncation_strategy}, "
f"for instance 'longest_first' or 'only_second'."
)
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)
overflowing_tokens = pair_ids[-window_len:]
overflowing_token_boxes = pair_token_boxes[-window_len:]
pair_ids = pair_ids[:-num_tokens_to_remove]
pair_token_boxes = pair_token_boxes[:-num_tokens_to_remove]
else:
logger.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,
token_boxes,
pair_ids,
pair_token_boxes,
labels,
overflowing_tokens,
overflowing_token_boxes,
overflowing_labels,
)
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,
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 self.padding_side:
- '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).
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
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)
if self.padding_side == "right":
if return_attention_mask:
encoded_inputs["attention_mask"] = encoded_inputs["attention_mask"] + [0] * difference
if "token_type_ids" in encoded_inputs:
encoded_inputs["token_type_ids"] = (
encoded_inputs["token_type_ids"] + [self.pad_token_type_id] * difference
)
if "bbox" in encoded_inputs:
encoded_inputs["bbox"] = encoded_inputs["bbox"] + [self.pad_token_box] * difference
if "labels" in encoded_inputs:
encoded_inputs["labels"] = encoded_inputs["labels"] + [self.pad_token_label] * difference
if "special_tokens_mask" in encoded_inputs:
encoded_inputs["special_tokens_mask"] = encoded_inputs["special_tokens_mask"] + [1] * difference
encoded_inputs[self.model_input_names[0]] = required_input + [self.pad_token_id] * difference
elif self.padding_side == "left":
if return_attention_mask:
encoded_inputs["attention_mask"] = [0] * difference + encoded_inputs["attention_mask"]
if "token_type_ids" in encoded_inputs:
encoded_inputs["token_type_ids"] = [self.pad_token_type_id] * difference + encoded_inputs[
"token_type_ids"
]
if "bbox" in encoded_inputs:
encoded_inputs["bbox"] = [self.pad_token_box] * difference + encoded_inputs["bbox"]
if "labels" in encoded_inputs:
encoded_inputs["labels"] = [self.pad_token_label] * difference + encoded_inputs["labels"]
if "special_tokens_mask" in encoded_inputs:
encoded_inputs["special_tokens_mask"] = [1] * difference + encoded_inputs["special_tokens_mask"]
encoded_inputs[self.model_input_names[0]] = [self.pad_token_id] * difference + required_input
else:
raise ValueError("Invalid padding strategy:" + str(self.padding_side))
return encoded_inputs