Remove slow tokenizer break pipeline loading.

[jiqing-feng]

System Info

latest main branch transformers

Who can help?

@itazap

Information

• The official example scripts
• My own modified scripts

Tasks

• An officially supported task in the examples folder (such as GLUE/SQuAD, ...)
• My own task or dataset (give details below)

Reproduction

from transformers import pipeline

pipeline("document-question-answering", model="impira/layoutlm-document-qa")

output:

Traceback (most recent call last):
  File "<stdin>", line 1, in <module>
  File "/home/jiqing/transformers/src/transformers/pipelines/__init__.py", line 1085, in pipeline
    return pipeline_class(model=model, task=task, **kwargs)
           ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/home/jiqing/transformers/src/transformers/pipelines/document_question_answering.py", line 150, in __init__
    raise ValueError(
ValueError: `DocumentQuestionAnsweringPipeline` requires a fast tokenizer, but a slow tokenizer (`RobertaTokenizer`) is provided.

Expected behavior

Regression PR #40936

cc @zucchini-nlp

[zucchini-nlp]

cc @itazap for tokenizers :)