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
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
|
/* eslint-disable */
var globRequire = require
console.log = (..._args: any[]) => {}
describe('quicktourExample', () => {
function require(mod: string) {
if (mod.startsWith('tokenizers')) {
return globRequire('../../')
} else {
return globRequire(mod)
}
}
it.skip('trains the tokenizer', async () => {
// START init_tokenizer
let { Tokenizer } = require('tokenizers')
let { BPE } = require('tokenizers')
let tokenizer = new Tokenizer(BPE.init({}, [], { unkToken: '[UNK]' }))
// END init_tokenizer
// START init_trainer
let { bpeTrainer } = require('tokenizers')
let trainer = bpeTrainer({
specialTokens: ['[UNK]', '[CLS]', '[SEP]', '[PAD]', '[MASK]'],
})
// END init_trainer
// START init_pretok
let { whitespacePreTokenizer } = require('tokenizers')
tokenizer.setPreTokenizer(whitespacePreTokenizer())
// END init_pretok
// START train
let files = ['test', 'train', 'valid'].map((split) => `data/wikitext-103-raw/wiki.${split}.raw`)
tokenizer.train(files, trainer)
// END train
// START save
tokenizer.save('data/tokenizer-wiki.json')
// END save
})
it('shows a quicktour example', async () => {
let { Tokenizer } = require('tokenizers')
// START reload_tokenizer
let tokenizer = Tokenizer.fromFile('data/tokenizer-wiki.json')
// END reload_tokenizer
// START encode
var output = await tokenizer.encode("Hello, y'all! How are you 😁 ?")
// END encode
// START print_tokens
console.log(output.getTokens())
// ["Hello", ",", "y", "'", "all", "!", "How", "are", "you", "[UNK]", "?"]
// END print_tokens
expect(output.getTokens()).toEqual(['Hello', ',', 'y', "'", 'all', '!', 'How', 'are', 'you', '[UNK]', '?'])
// START print_ids
console.log(output.getIds())
// [27253, 16, 93, 11, 5097, 5, 7961, 5112, 6218, 0, 35]
// END print_ids
expect(output.getIds()).toEqual([27253, 16, 93, 11, 5097, 5, 7961, 5112, 6218, 0, 35])
// START print_offsets
let offsets = output.getOffsets()
console.log(offsets[9])
// (26, 27)
// END print_offsets
expect(offsets[9]).toEqual([26, 27])
// START use_offsets
let { slice } = require('tokenizers')
let sentence = "Hello, y'all! How are you 😁 ?"
let [start, end] = offsets[9]
console.log(slice(sentence, start, end))
// "😁"
// END use_offsets
expect(slice(sentence, start, end)).toEqual('😁')
// START check_sep
console.log(tokenizer.tokenToId('[SEP]'))
// 2
// END check_sep
expect(tokenizer.tokenToId('[SEP]')).toEqual(2)
// START init_template_processing
let { templateProcessing } = require('tokenizers')
tokenizer.setPostProcessor(
templateProcessing('[CLS] $A [SEP]', '[CLS] $A [SEP] $B:1 [SEP]:1', [
['[CLS]', tokenizer.tokenToId('[CLS]')],
['[SEP]', tokenizer.tokenToId('[SEP]')],
]),
)
// END init_template_processing
// START print_special_tokens
var output = await tokenizer.encode("Hello, y'all! How are you 😁 ?")
console.log(output.getTokens())
// ["[CLS]", "Hello", ",", "y", "'", "all", "!", "How", "are", "you", "[UNK]", "?", "[SEP]"]
// END print_special_tokens
expect(output.getTokens()).toEqual([
'[CLS]',
'Hello',
',',
'y',
"'",
'all',
'!',
'How',
'are',
'you',
'[UNK]',
'?',
'[SEP]',
])
// START print_special_tokens_pair
var output = await tokenizer.encode("Hello, y'all!", 'How are you 😁 ?')
console.log(output.getTokens())
// ["[CLS]", "Hello", ",", "y", "'", "all", "!", "[SEP]", "How", "are", "you", "[UNK]", "?", "[SEP]"]
// END print_special_tokens_pair
expect(output.getTokens()).toEqual([
'[CLS]',
'Hello',
',',
'y',
"'",
'all',
'!',
'[SEP]',
'How',
'are',
'you',
'[UNK]',
'?',
'[SEP]',
])
// START print_type_ids
console.log(output.getTypeIds())
// [0, 0, 0, 0, 0, 0, 0, 0, 1, 1, 1, 1, 1, 1]
// END print_type_ids
expect(output.getTypeIds()).toEqual([0, 0, 0, 0, 0, 0, 0, 0, 1, 1, 1, 1, 1, 1])
// START encode_batch
var output = await tokenizer.encodeBatch(["Hello, y'all!", 'How are you 😁 ?'])
// END encode_batch
// START encode_batch_pair
// var output = await tokenizer.encodeBatch(
// [["Hello, y'all!", "How are you 😁 ?"], ["Hello to you too!", "I'm fine, thank you!"]]
// );
// END encode_batch_pair
// START enable_padding
tokenizer.setPadding({ padId: 3, padToken: '[PAD]' })
// END enable_padding
// START print_batch_tokens
var output = await tokenizer.encodeBatch(["Hello, y'all!", 'How are you 😁 ?'])
console.log(output[1].getTokens())
// ["[CLS]", "How", "are", "you", "[UNK]", "?", "[SEP]", "[PAD]"]
// END print_batch_tokens
expect(output[1].getTokens()).toEqual(['[CLS]', 'How', 'are', 'you', '[UNK]', '?', '[SEP]', '[PAD]'])
// START print_attention_mask
console.log(output[1].getAttentionMask())
// [1, 1, 1, 1, 1, 1, 1, 0]
// END print_attention_mask
expect(output[1].getAttentionMask()).toEqual([1, 1, 1, 1, 1, 1, 1, 0])
})
})
|
