Week 1 lab - ReadTimeoutError only for model (not datasets)

I am able to load the datasets and print the human generated summaries but the model (FLAN-T5) fails to load with ReadTimeoutError. Specifically the cell that includes these lines:

model_name=‘google/flan-t5-base’
model = AutoModelForSeq2SeqLM.from_pretrained(model_name)

In the FAQ there was a mention of huggingface servers issues but then I’d expect the dataset to also not load. So that can’t be it.

Would appreciate your help!

7 Likes

having the exact same issue here. Having the Timeout error only for model loading.

1 Like

Facing the same issue here. Thank you for bringing this up. Not sure where I had to bring up this issue.

Same issue, Timeout when build_dataset after loading FLAN-T5 Model Fine-Tuned with Summarization Instruction

Same issue
Just to experiment tried smaller model “google/flan-t5-small”, but problem also the same

1 Like

Same issue here, do we just wait and rerun the notebook.

1 Like

Same issue for Lab 2 when try to use the original model. Hope can get solutions.

I am also seeing the same issue.

model_name='google/flan-t5-base'

model = AutoModelForSeq2SeqLM.from_pretrained(model_name)
---------------------------------------------------------------------------
TimeoutError                              Traceback (most recent call last)
File /opt/conda/lib/python3.11/site-packages/urllib3/connectionpool.py:404, in HTTPConnectionPool._make_request(self, conn, method, url, timeout, chunked, **httplib_request_kw)
    403 try:
--> 404     self._validate_conn(conn)
    405 except (SocketTimeout, BaseSSLError) as e:
    406     # Py2 raises this as a BaseSSLError, Py3 raises it as socket timeout.

File /opt/conda/lib/python3.11/site-packages/urllib3/connectionpool.py:1060, in HTTPSConnectionPool._validate_conn(self, conn)
   1059 if not getattr(conn, "sock", None):  # AppEngine might not have  `.sock`
-> 1060     conn.connect()
   1062 if not conn.is_verified:

File /opt/conda/lib/python3.11/site-packages/urllib3/connection.py:419, in HTTPSConnection.connect(self)
    417     context.load_default_certs()
--> 419 self.sock = ssl_wrap_socket(
    420     sock=conn,
    421     keyfile=self.key_file,
    422     certfile=self.cert_file,
    423     key_password=self.key_password,
    424     ca_certs=self.ca_certs,
    425     ca_cert_dir=self.ca_cert_dir,
    426     ca_cert_data=self.ca_cert_data,
    427     server_hostname=server_hostname,
    428     ssl_context=context,
    429     tls_in_tls=tls_in_tls,
    430 )
    432 # If we're using all defaults and the connection
    433 # is TLSv1 or TLSv1.1 we throw a DeprecationWarning
    434 # for the host.

File /opt/conda/lib/python3.11/site-packages/urllib3/util/ssl_.py:449, in ssl_wrap_socket(sock, keyfile, certfile, cert_reqs, ca_certs, server_hostname, ssl_version, ciphers, ssl_context, ca_cert_dir, key_password, ca_cert_data, tls_in_tls)
    448 if send_sni:
--> 449     ssl_sock = _ssl_wrap_socket_impl(
    450         sock, context, tls_in_tls, server_hostname=server_hostname
    451     )
    452 else:

1 Like

Facing issue for Flan-T5 in Week 2 lab
Fine-Tune a Generative AI Model for Dialogue Summarization also.

Same here.

Hi everyone! Thank you for reporting! We will look into this and update this thread asap.

3 Likes

Same error

Hi everyone! Just a friendly note. If you’re encountering the same issue, you can watch the topic instead of replying with a short note. That will help avoid flooding the thread so posts with more info will be more visible. Watching the thread will enable notifications as soon as a new reply is posted. You can select this at the bottom left of this page as shown in the screenshot below. Thank you!

2 Likes

would voting up be a good option? so the poster can tell if others are having similar issues? or should we look at whether people are watching the topic?

1 Like

Just wait for the issue to be resolved.

Additional feedback is not necessary.

1 Like

Same Issue

1 Like

Thank you! I have the course deadline tomorrow, do you think it will be solved by then? if not will the access be extended?
Thank you!
Best

1 Like

Hello Chris, my deadline is tomorrow (I was about to conclude the course Today).

Could you please extend the deadline, so I can finish the course as expected (when you fix it) and get my certificate?

Thanks in advance.

1 Like

Hi Afonso. Please contact Coursera (instructions here) to know if an extension is possible for your account. You can mention that there is a bug in the course and the dev team is still working on it. You can let me know the response you got. Thank you, and sorry for the trouble!

Hi Victor! Please refer to the response here. Please let me know how it goes. Thank you!

1 Like