CSS-LM

1
# coding=utf-8
2
# Copyright 2018 The HuggingFace Inc. team.
3
#
4
# Licensed under the Apache License, Version 2.0 (the "License");
5
# you may not use this file except in compliance with the License.
6
# You may obtain a copy of the License at
7
#
8
#     http://www.apache.org/licenses/LICENSE-2.0
9
#
10
# Unless required by applicable law or agreed to in writing, software
11
# distributed under the License is distributed on an "AS IS" BASIS,
12
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
13
# See the License for the specific language governing permissions and
14
# limitations under the License.
15
"""Tokenization classes for DistilBERT."""
16

17

18
import logging
19

20
from .tokenization_bert import BertTokenizer, BertTokenizerFast
21

22

23
logger = logging.getLogger(__name__)
24

25
VOCAB_FILES_NAMES = {"vocab_file": "vocab.txt"}
26

27
PRETRAINED_VOCAB_FILES_MAP = {
28
    "vocab_file": {
29
        "distilbert-base-uncased": "https://s3.amazonaws.com/models.huggingface.co/bert/bert-base-uncased-vocab.txt",
30
        "distilbert-base-uncased-distilled-squad": "https://s3.amazonaws.com/models.huggingface.co/bert/bert-large-uncased-vocab.txt",
31
        "distilbert-base-cased": "https://s3.amazonaws.com/models.huggingface.co/bert/bert-base-cased-vocab.txt",
32
        "distilbert-base-cased-distilled-squad": "https://s3.amazonaws.com/models.huggingface.co/bert/bert-large-cased-vocab.txt",
33
        "distilbert-base-german-cased": "https://s3.amazonaws.com/models.huggingface.co/bert/distilbert-base-german-cased-vocab.txt",
34
        "distilbert-base-multilingual-cased": "https://s3.amazonaws.com/models.huggingface.co/bert/bert-base-multilingual-cased-vocab.txt",
35
    }
36
}
37

38
PRETRAINED_POSITIONAL_EMBEDDINGS_SIZES = {
39
    "distilbert-base-uncased": 512,
40
    "distilbert-base-uncased-distilled-squad": 512,
41
    "distilbert-base-cased": 512,
42
    "distilbert-base-cased-distilled-squad": 512,
43
    "distilbert-base-german-cased": 512,
44
    "distilbert-base-multilingual-cased": 512,
45
}
46

47

48
PRETRAINED_INIT_CONFIGURATION = {
49
    "distilbert-base-uncased": {"do_lower_case": True},
50
    "distilbert-base-uncased-distilled-squad": {"do_lower_case": True},
51
    "distilbert-base-cased": {"do_lower_case": False},
52
    "distilbert-base-cased-distilled-squad": {"do_lower_case": False},
53
    "distilbert-base-german-cased": {"do_lower_case": False},
54
    "distilbert-base-multilingual-cased": {"do_lower_case": False},
55
}
56

57

58
class DistilBertTokenizer(BertTokenizer):
59
    r"""
60
    Constructs a  DistilBertTokenizer.
61

62
    :class:`~transformers.DistilBertTokenizer is identical to :class:`~transformers.BertTokenizer` and runs end-to-end
63
    tokenization: punctuation splitting + wordpiece.
64

65
    Refer to superclass :class:`~transformers.BertTokenizer` for usage examples and documentation concerning
66
    parameters.
67
    """
68

69
    vocab_files_names = VOCAB_FILES_NAMES
70
    pretrained_vocab_files_map = PRETRAINED_VOCAB_FILES_MAP
71
    max_model_input_sizes = PRETRAINED_POSITIONAL_EMBEDDINGS_SIZES
72
    pretrained_init_configuration = PRETRAINED_INIT_CONFIGURATION
73
    model_input_names = ["attention_mask"]
74

75

76
class DistilBertTokenizerFast(BertTokenizerFast):
77
    r"""
78
    Constructs a  "Fast" DistilBertTokenizer (backed by HuggingFace's `tokenizers` library).
79

80
    :class:`~transformers.DistilBertTokenizerFast` is identical to :class:`~transformers.BertTokenizerFast` and runs end-to-end
81
    tokenization: punctuation splitting + wordpiece.
82

83
    Refer to superclass :class:`~transformers.BertTokenizerFast` for usage examples and documentation concerning
84
    parameters.
85
    """
86

87
    vocab_files_names = VOCAB_FILES_NAMES
88
    pretrained_vocab_files_map = PRETRAINED_VOCAB_FILES_MAP
89
    max_model_input_sizes = PRETRAINED_POSITIONAL_EMBEDDINGS_SIZES
90
    pretrained_init_configuration = PRETRAINED_INIT_CONFIGURATION
91
    model_input_names = ["attention_mask"]
92