Update pick_best_tokenizer.py
Browse files- pick_best_tokenizer.py +1 -0
pick_best_tokenizer.py
CHANGED
|
@@ -12,6 +12,7 @@ class PickBestTokenizer(PreTrainedTokenizer):
|
|
| 12 |
self.tokenizers_offsets = []
|
| 13 |
self.vocab = {}
|
| 14 |
self._vocab_size = sum(len(tokenizer) for tokenizer in self.tokenizers)
|
|
|
|
| 15 |
|
| 16 |
offset = 0
|
| 17 |
for i, tokenizer in enumerate(self.tokenizers):
|
|
|
|
| 12 |
self.tokenizers_offsets = []
|
| 13 |
self.vocab = {}
|
| 14 |
self._vocab_size = sum(len(tokenizer) for tokenizer in self.tokenizers)
|
| 15 |
+
self.pad_token = '[0]'+self.tokenizers[0].pad_token
|
| 16 |
|
| 17 |
offset = 0
|
| 18 |
for i, tokenizer in enumerate(self.tokenizers):
|