(self)
| 22 | """Mock tokenizer for testing.""" |
| 23 | |
| 24 | def __init__(self): |
| 25 | self._vocab = {"[PAD]": 0, "[UNK]": 1, "[CLS]": 2, "[SEP]": 3} |
| 26 | self._vocab_size = 1000 |
| 27 | self._pad_token = "[PAD]" |
| 28 | self._pad_token_id = 0 |
| 29 | self.model_max_length = 512 |
| 30 | self.padding_side = "right" |
| 31 | self.truncation_side = "right" |
| 32 | super().__init__() |
| 33 | |
| 34 | def encode(self, text, **kwargs): |
| 35 | return [1, 2, 3] # Mock token IDs |