fix(tokenizer): 更新特殊token定义以支持新的对话格式
This commit is contained in:
@@ -8,7 +8,7 @@ from typing import List, Union
|
||||
class BpeTokenizer:
|
||||
def __init__(self, path=None):
|
||||
self._control_tokens = ["<bos>", "<eos>", "<pad>"]
|
||||
self._special_tokens = ["<|user|>", "<|system|>"]
|
||||
self._special_tokens = ["<|im_start|>", "<|im_end|>"]
|
||||
model = BPE()
|
||||
tokenizer = Tokenizer(model)
|
||||
tokenizer.normalizer = normalizers.Sequence([
|
||||
|
||||
Reference in New Issue
Block a user