master
/ transformers / models / longformer / tokenization_longformer_fast.py

tokenization_longformer_fast.py @3c11360 

69624d8
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
# coding=utf-8
# Copyright 2020 The Allen Institute for AI team and 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.

from ...utils import logging
from ..roberta.tokenization_roberta_fast import RobertaTokenizerFast
from .tokenization_longformer import LongformerTokenizer


logger = logging.get_logger(__name__)


VOCAB_FILES_NAMES = {"vocab_file": "vocab.json", "merges_file": "merges.txt", "tokenizer_file": "tokenizer.json"}

PRETRAINED_VOCAB_FILES_MAP = {
    "vocab_file": {
        "allenai/longformer-base-4096": "https://huggingface.co/allenai/longformer-base-4096/resolve/main/vocab.json",
        "allenai/longformer-large-4096": "https://huggingface.co/allenai/longformer-large-4096/resolve/main/vocab.json",
        "allenai/longformer-large-4096-finetuned-triviaqa": "https://huggingface.co/allenai/longformer-large-4096-finetuned-triviaqa/resolve/main/vocab.json",
        "allenai/longformer-base-4096-extra.pos.embd.only": "https://huggingface.co/allenai/longformer-base-4096-extra.pos.embd.only/resolve/main/vocab.json",
        "allenai/longformer-large-4096-extra.pos.embd.only": "https://huggingface.co/allenai/longformer-large-4096-extra.pos.embd.only/resolve/main/vocab.json",
    },
    "merges_file": {
        "allenai/longformer-base-4096": "https://huggingface.co/allenai/longformer-base-4096/resolve/main/merges.txt",
        "allenai/longformer-large-4096": "https://huggingface.co/allenai/longformer-large-4096/resolve/main/merges.txt",
        "allenai/longformer-large-4096-finetuned-triviaqa": "https://huggingface.co/allenai/longformer-large-4096-finetuned-triviaqa/resolve/main/merges.txt",
        "allenai/longformer-base-4096-extra.pos.embd.only": "https://huggingface.co/allenai/longformer-base-4096-extra.pos.embd.only/resolve/main/merges.txt",
        "allenai/longformer-large-4096-extra.pos.embd.only": "https://huggingface.co/allenai/longformer-large-4096-extra.pos.embd.only/resolve/main/merges.txt",
    },
    "tokenizer_file": {
        "allenai/longformer-base-4096": "https://huggingface.co/allenai/longformer-base-4096/resolve/main/tokenizer.json",
        "allenai/longformer-large-4096": "https://huggingface.co/allenai/longformer-large-4096/resolve/main/tokenizer.json",
        "allenai/longformer-large-4096-finetuned-triviaqa": "https://huggingface.co/allenai/longformer-large-4096-finetuned-triviaqa/resolve/main/tokenizer.json",
        "allenai/longformer-base-4096-extra.pos.embd.only": "https://huggingface.co/allenai/longformer-base-4096-extra.pos.embd.only/resolve/main/tokenizer.json",
        "allenai/longformer-large-4096-extra.pos.embd.only": "https://huggingface.co/allenai/longformer-large-4096-extra.pos.embd.only/resolve/main/tokenizer.json",
    },
}

PRETRAINED_POSITIONAL_EMBEDDINGS_SIZES = {
    "allenai/longformer-base-4096": 4096,
    "allenai/longformer-large-4096": 4096,
    "allenai/longformer-large-4096-finetuned-triviaqa": 4096,
    "allenai/longformer-base-4096-extra.pos.embd.only": 4096,
    "allenai/longformer-large-4096-extra.pos.embd.only": 4096,
}


class LongformerTokenizerFast(RobertaTokenizerFast):
    r"""
    Construct a "fast" Longformer tokenizer (backed by HuggingFace's *tokenizers* library).

    [`LongformerTokenizerFast`] is identical to [`RobertaTokenizerFast`]. Refer
    to the superclass for usage examples and documentation concerning parameters.
    """
    # merges and vocab same as Roberta
    vocab_files_names = VOCAB_FILES_NAMES
    pretrained_vocab_files_map = PRETRAINED_VOCAB_FILES_MAP
    max_model_input_sizes = PRETRAINED_POSITIONAL_EMBEDDINGS_SIZES
    slow_tokenizer_class = LongformerTokenizer