Connection error when building


I tried updating our gadio implementation in our HF Space, to better handle the increased traffic we have had the last few days. The build seems to be successful but we get a connection error when loading the tokenizer (although I guess that part is anecdotal and the connection error might be more general?). Here is the full traceback:

Traceback (most recent call last):
  File "", line 19, in <module>
    tokenizer = AutoTokenizer.from_pretrained('gpt2')
  File "/home/user/.local/lib/python3.8/site-packages/transformers/models/auto/", line 464, in from_pretrained
    tokenizer_config = get_tokenizer_config(pretrained_model_name_or_path, **kwargs)
  File "/home/user/.local/lib/python3.8/site-packages/transformers/models/auto/", line 330, in get_tokenizer_config
    resolved_config_file = cached_path(
  File "/home/user/.local/lib/python3.8/site-packages/transformers/", line 1491, in cached_path
    output_path = get_from_cache(
  File "/home/user/.local/lib/python3.8/site-packages/transformers/", line 1715, in get_from_cache
    raise ValueError(
ValueError: Connection error, and we cannot find the requested files in the cached path. Please try again or make sure your Internet connection is on.

Any ideas what we can do to fix this?

Thanks in advance,

Hi Theodore!

This same as the same issue as the one discussed in Build error : failed to create endpoint hungry_leavitt on network bridge: adding interface veth1291603 to bridge docker0 failed: exchange full - #2 by mikeee. Looking into it!


Thank you! I’ll keep an eye for updates.

So the build is successful now! However, the HF space page and the site page are showing a 504 Gateway Time-out. The log says has a Running on local URL message as well, if that helps.

Edit: rebuilding to update some gradio deprecated options gave the same connection time out as before.