File size: 5,948 Bytes
ed7a497
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
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
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
# coding=utf-8
# Copyright 2022 The HuggingFace Team. 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 unittest

from datasets import load_dataset

from transformers import BloomTokenizerFast
from transformers.testing_utils import require_tokenizers

from ...test_tokenization_common import TokenizerTesterMixin


@require_tokenizers
class BloomTokenizationTest(TokenizerTesterMixin, unittest.TestCase):
    slow_tokenizer_class = None
    rust_tokenizer_class = BloomTokenizerFast
    tokenizer_class = BloomTokenizerFast
    test_rust_tokenizer = True
    test_slow_tokenizer = False
    from_pretrained_vocab_key = "tokenizer_file"
    special_tokens_map = {"bos_token": "<s>", "eos_token": "</s>", "unk_token": "<unk>", "pad_token": "<pad>"}

    def setUp(self):
        super().setUp()
        tokenizer = BloomTokenizerFast.from_pretrained("bigscience/tokenizer")
        tokenizer.save_pretrained(self.tmpdirname)

    def get_rust_tokenizer(self, **kwargs):
        kwargs.update(self.special_tokens_map)
        return BloomTokenizerFast.from_pretrained(self.tmpdirname, **kwargs)

    def test_encodings_from_sample_data(self):
        """
        Assert that the created tokens are the same than the hard-coded ones
        """
        tokenizer = self.get_rust_tokenizer()

        INPUT_SENTENCES = ["The quick brown fox</s>", "jumps over the lazy dog</s>"]
        TARGET_TOKENS = [[2175, 23714, 73173, 144252, 2], [77, 132619, 3478, 368, 109586, 35433, 2]]

        computed_tokens = tokenizer.batch_encode_plus(INPUT_SENTENCES)["input_ids"]
        self.assertListEqual(TARGET_TOKENS, computed_tokens)

        decoded_tokens = tokenizer.batch_decode(computed_tokens)
        self.assertListEqual(decoded_tokens, INPUT_SENTENCES)

    def test_padding(self, max_length=6):
        for tokenizer, pretrained_name, kwargs in self.tokenizers_list:
            with self.subTest(f"{tokenizer.__class__.__name__} ({pretrained_name})"):
                tokenizer_r = self.rust_tokenizer_class.from_pretrained(pretrained_name, **kwargs)
                # tokenizer_r.pad_token = None # Hotfixing padding = None
                # Simple input
                s = "This is a simple input"
                s2 = ["This is a simple input 1", "This is a simple input 2"]
                p = ("This is a simple input", "This is a pair")
                p2 = [
                    ("This is a simple input 1", "This is a simple input 2"),
                    ("This is a simple pair 1", "This is a simple pair 2"),
                ]

                # Simple input tests
                try:
                    tokenizer_r.encode(s, max_length=max_length)
                    tokenizer_r.encode_plus(s, max_length=max_length)

                    tokenizer_r.batch_encode_plus(s2, max_length=max_length)
                    tokenizer_r.encode(p, max_length=max_length)
                    tokenizer_r.batch_encode_plus(p2, max_length=max_length)
                except ValueError:
                    self.fail("Bloom Tokenizer should be able to deal with padding")

                tokenizer_r.pad_token = None  # Hotfixing padding = None
                self.assertRaises(ValueError, tokenizer_r.encode, s, max_length=max_length, padding="max_length")

                # Simple input
                self.assertRaises(ValueError, tokenizer_r.encode_plus, s, max_length=max_length, padding="max_length")

                # Simple input
                self.assertRaises(
                    ValueError,
                    tokenizer_r.batch_encode_plus,
                    s2,
                    max_length=max_length,
                    padding="max_length",
                )

                # Pair input
                self.assertRaises(ValueError, tokenizer_r.encode, p, max_length=max_length, padding="max_length")

                # Pair input
                self.assertRaises(ValueError, tokenizer_r.encode_plus, p, max_length=max_length, padding="max_length")

                # Pair input
                self.assertRaises(
                    ValueError,
                    tokenizer_r.batch_encode_plus,
                    p2,
                    max_length=max_length,
                    padding="max_length",
                )

    def test_encodings_from_xnli_dataset(self):
        """
        Tests the tokenizer downloaded from here:
            - https://huggingface.co/bigscience/tokenizer/
        """
        tokenizer = self.get_rust_tokenizer()
        ds = load_dataset("xnli", "all_languages", split="test", streaming=True)

        sample_data = next(iter(ds))["premise"]  # pick up one data
        input_text = list(sample_data.values())

        output_tokens = list(map(tokenizer.encode, input_text))
        predicted_text = [tokenizer.decode(x, clean_up_tokenization_spaces=False) for x in output_tokens]
        self.assertListEqual(predicted_text, input_text)

    def test_pretrained_model_lists(self):
        # The test has to be overriden because BLOOM uses ALiBi positional embeddings that does not have
        # any sequence length constraints. This test of the parent class will fail since it relies on the
        # maximum sequence length of the positoonal embeddings.
        self.assertGreaterEqual(len(self.tokenizer_class.pretrained_vocab_files_map), 1)
        self.assertGreaterEqual(len(list(self.tokenizer_class.pretrained_vocab_files_map.values())[0]), 1)