"AutoTokenizer.from_pretrained" does not work when loading a pretrained Albert model

[iamfaith]

Environment info

• transformers version:
• Platform: 5.4.0-53-generic not good when I use BERT for seq2seq model in keyphrase generation #59~18.04.1-Ubuntu SMP Wed Oct 21 12:14:56 UTC 2020 x86_64 x86_64 x86_64 GNU/Linux
• Python version: 3.7.9
• PyTorch version (GPU?): 1.7.0
• Tensorflow version (GPU?): N/A
• Using GPU in script?: No
• Using distributed or parallel set-up in script?: No

Who can help

Information

Model I am using (Bert, XLNet ...):

The problem arises when using:

• the official example scripts: (give details below)
• my own modified scripts: (give details below)

The tasks I am working on is:

• an official GLUE/SQUaD task: (give the name)
• my own task or dataset: (give details below)

To reproduce

Steps to reproduce the behavior:

1. Install PyTorch from the official website as well as the transformers via pip.
2. Using the following pre-trained model:

from transformers import AutoTokenizer, AutoModelForMaskedLM

tokenizer = AutoTokenizer.from_pretrained("ckiplab/albert-tiny-chinese")

model = AutoModelForMaskedLM.from_pretrained("ckiplab/albert-tiny-chinese")

3. Error:

Downloading: 100%|███████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████| 683/683 [00:00<00:00, 1.32MB/s]
Downloading: 100%|████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████| 112/112 [00:00<00:00, 215kB/s]
Downloading: 100%|████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████| 174/174 [00:00<00:00, 334kB/s]
Traceback (most recent call last):
  File "/home/faith/torch_tutorials/torch_chatbot.py", line 30, in <module>
    tokenizer = AutoTokenizer.from_pretrained("ckiplab/albert-tiny-chinese")
  File "/home/faith/miniconda3/envs/torch/lib/python3.7/site-packages/transformers/tokenization_auto.py", line 341, in from_pretrained
    return tokenizer_class_py.from_pretrained(pretrained_model_name_or_path, *inputs, **kwargs)
  File "/home/faith/miniconda3/envs/torch/lib/python3.7/site-packages/transformers/tokenization_utils_base.py", line 1653, in from_pretrained
    resolved_vocab_files, pretrained_model_name_or_path, init_configuration, *init_inputs, **kwargs
  File "/home/faith/miniconda3/envs/torch/lib/python3.7/site-packages/transformers/tokenization_utils_base.py", line 1725, in _from_pretrained
    tokenizer = cls(*init_inputs, **init_kwargs)
  File "/home/faith/miniconda3/envs/torch/lib/python3.7/site-packages/transformers/tokenization_albert.py", line 149, in __init__
    self.sp_model.Load(vocab_file)
  File "/home/faith/miniconda3/envs/torch/lib/python3.7/site-packages/sentencepiece.py", line 367, in Load
    return self.LoadFromFile(model_file)
  File "/home/faith/miniconda3/envs/torch/lib/python3.7/site-packages/sentencepiece.py", line 177, in LoadFromFile
    return _sentencepiece.SentencePieceProcessor_LoadFromFile(self, arg)
TypeError: not a string

Expected behavior

Expect to download this model correctly with error prompting.

[thomwolf]

Can you share your version of transformers, tokenizers?

[NielsRogge]

I can reproduce this in a Colab notebook when doing pip install transformers.

• Transformers version 3.5.1
• Tokenizers version 0.9.3

Might be solved with v4?

[akar5h]

I am having the same issue with AlbertTokenizer.from_pretrained

[github-actions]

This issue has been automatically marked as stale and been closed because it has not had recent activity. Thank you for your contributions.

If you think this still needs to be addressed please comment on this thread.

[wenHK]

i have the same question!