File size: 1,502 Bytes
c766a08
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
293bad6
 
c766a08
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
"""

## 默认 use_fast=True 报错
  File "E:\workspace\common\vocab-zoo\tokenizer-arena\utils\zh_util.py", line 190, in <module>
    print(iter_vocab(tokenizer, name=name))
  File "E:\workspace\common\vocab-zoo\tokenizer-arena\utils\zh_util.py", line 144, in iter_vocab
    dist_length, mean_length = get_coding_length(tokenizer, zh_tokens, filter=lambda k: not is_zh_char(k))
  File "E:\workspace\common\vocab-zoo\tokenizer-arena\utils\zh_util.py", line 34, in get_coding_length
    tokens = tokenizer.encode(word)
  File "C:\Users\xusong\Miniconda3\envs\py3.10-torch1.13-hf.latest\lib\site-packages\transformers\tokenization_utils_base.py", line 2600, in encode
    encoded_inputs = self.encode_plus(
  File "C:\Users\xusong\Miniconda3\envs\py3.10-torch1.13-hf.latest\lib\site-packages\transformers\tokenization_utils_base.py", line 3008, in encode_plus
    return self._encode_plus(
  File "C:\Users\xusong\Miniconda3\envs\py3.10-torch1.13-hf.latest\lib\site-packages\transformers\tokenization_utils_fast.py", line 576, in _encode_plus
    batched_output = self._batch_encode_plus(
  File "C:\Users\xusong\Miniconda3\envs\py3.10-torch1.13-hf.latest\lib\site-packages\transformers\tokenization_utils_fast.py", line 504, in _batch_encode_plus
    encodings = self._tokenizer.encode_batch(
pyo3_runtime.PanicException: AddedVocabulary bad split
"""

from transformers import AutoTokenizer

tokenizer = AutoTokenizer.from_pretrained("lmsys/fastchat-t5-3b-v1.0", trust_remote_code=True, use_fast=False)