🐛 Bug

No information from the package on SSL error encountered, making it difficult to troubleshoot or figure out a workaround

Information

When trying to do:
TFAutoModelWithLMHead.from_pretrained("t5-small")

Get an error:
TypeError: stat: path should be string, bytes, os.PathLike or integer, not NoneType

The above is a result of an SSL error encountered when trying to fetch the model, however, since the exception handling isn't proper within file_utils.py I don't come to know of it, unless I debug.

Model I am using (Bert, XLNet ...): T5

Language I am using the model on (English, Chinese ...): English

The problem arises when using:

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

https://huggingface.co/transformers/usage.html#summarization

The tasks I am working on is:

• an official GLUE/SQUaD task: (give the name)
• [ x] my own task or dataset: (give details below)
  Just getting familiar with transformers for summarization

To reproduce

You need a machine with an expired certificate for proxy etc.

Steps to reproduce the behavior:

1. See information above

Expected behavior

If there has been an issue in fetching the pre-trained model from s3 bucket etc. I should get an error to that effect.

Environment info

• transformers version: 2.10.0
• Platform: Windows 10
• Python version: 3.7.4
• PyTorch version (GPU?): NA
• Tensorflow version (GPU?): 2.0.0 (Yes)
• Using GPU in script?: No
• Using distributed or parallel set-up in script?: No