File size: 5,231 Bytes
adc0ea3 | 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 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 | """
Simple Word-Level Tokenizer
==============================
A basic tokenizer for demonstration purposes.
Converts text to token IDs with special tokens.
"""
import re
import json
from typing import Dict, List, Optional
from collections import Counter
from tqdm import tqdm
class SimpleTokenizer:
"""
A simple word-level tokenizer with special tokens.
Special Tokens:
- [PAD]: Padding token (id=0)
- [UNK]: Unknown token (id=1)
- [CLS]: Classification token (id=2)
- [SEP]: Separator token (id=3)
"""
def __init__(self, vocab_size: int = 30000):
self.vocab_size = vocab_size
# Special tokens
self.special_tokens = {
'[PAD]': 0,
'[UNK]': 1,
'[CLS]': 2,
'[SEP]': 3,
}
# Word to ID mapping
self.word_to_id: Dict[str, int] = dict(self.special_tokens)
self.id_to_word: Dict[int, str] = {v: k for k, v in self.special_tokens.items()}
# Special token IDs
self.pad_token_id = 0
self.unk_token_id = 1
self.cls_token_id = 2
self.sep_token_id = 3
def _tokenize(self, text: str) -> List[str]:
"""
Split text into tokens (simple word-level tokenization).
Args:
text: Input text string
Returns:
List of tokens
"""
# Lowercase and basic cleaning
text = text.lower().strip()
# Simple word tokenization with punctuation handling
tokens = re.findall(r'\b\w+\b|[^\w\s]', text)
return tokens
def build_vocab(self, texts: List[str], min_freq: int = 2):
"""
Build vocabulary from a list of texts.
Args:
texts: List of text strings
min_freq: Minimum frequency for a word to be included
"""
# Count word frequencies
word_counts = Counter()
for text in tqdm(texts, desc="Building vocabulary"):
tokens = self._tokenize(text)
word_counts.update(tokens)
# Sort by frequency and take top vocab_size - special_tokens
max_words = self.vocab_size - len(self.special_tokens)
sorted_words = sorted(
word_counts.items(),
key=lambda x: x[1],
reverse=True
)
# Add words to vocabulary
for word, count in sorted_words[:max_words]:
if count >= min_freq and word not in self.word_to_id:
idx = len(self.word_to_id)
self.word_to_id[word] = idx
self.id_to_word[idx] = word
print(f"Vocabulary size: {len(self.word_to_id)}")
def encode(self, text: str, max_length: int = 128) -> Dict:
"""
Encode text to token IDs with attention mask.
Args:
text: Input text string
max_length: Maximum sequence length
Returns:
Dictionary with 'input_ids' and 'attention_mask' tensors
"""
import torch
# Tokenize
tokens = self._tokenize(text)
# Convert to IDs (with CLS and SEP)
token_ids = [self.cls_token_id]
for token in tokens[:max_length - 2]: # Reserve space for CLS and SEP
token_id = self.word_to_id.get(token, self.unk_token_id)
token_ids.append(token_id)
token_ids.append(self.sep_token_id)
# Create attention mask
attention_mask = [1] * len(token_ids)
# Pad to max_length
padding_length = max_length - len(token_ids)
token_ids.extend([self.pad_token_id] * padding_length)
attention_mask.extend([0] * padding_length)
return {
'input_ids': torch.tensor(token_ids, dtype=torch.long),
'attention_mask': torch.tensor(attention_mask, dtype=torch.long)
}
def decode(self, token_ids: List[int]) -> str:
"""
Decode token IDs back to text.
Args:
token_ids: List of token IDs
Returns:
Decoded text string
"""
tokens = []
for idx in token_ids:
if idx in [self.pad_token_id, self.cls_token_id, self.sep_token_id]:
continue
token = self.id_to_word.get(idx, '[UNK]')
tokens.append(token)
return ' '.join(tokens)
def save(self, path: str):
"""Save tokenizer vocabulary to JSON file."""
data = {
'vocab_size': self.vocab_size,
'word_to_id': self.word_to_id,
}
with open(path, 'w') as f:
json.dump(data, f, indent=2)
def load(self, path: str):
"""Load tokenizer vocabulary from JSON file."""
with open(path, 'r') as f:
data = json.load(f)
self.vocab_size = data['vocab_size']
self.word_to_id = data['word_to_id']
self.id_to_word = {int(v): k for k, v in self.word_to_id.items()}
def __len__(self) -> int:
return len(self.word_to_id)
|