Vision: Downloading pretrained models with torchvision gives HTTP Error 403

[mcleonard]

I'm getting a 403 error when I try to download pre-trained weights with torchvision 0.5.

torchvision.models.mobilenet_v2(pretrained=True) results in HTTPError: HTTP Error 403: Forbidden. I've tried multiple models on multiple machines on multiple networks, always get a 403 error.

I can download the model directly from the URL with wget. But for some reason downloading it through torchvision doesn't work. I can progress since I can download the weights, but it's unfortunate I have to load them in manually.

[CMobley7]

I can confirm. This issue started happening this afternoon. I don't think it has anything to do with the library but rather with where the models are hosted.

[uloma07]

Also having this is issue, when trying to download a pretrained vgg model. I am using Google colab.

Downloading: "https://download.pytorch.org/models/vgg19-dcbb9e9d.pth" to /root/.cache/torch/checkpoints/vgg19-dcbb9e9d.pth

HTTPError Traceback (most recent call last)
in ()

----> vgg_model = models.vgg19(pretrained=True)

/usr/lib/python3.6/urllib/request.py in http_error_default(self, req, fp, code, msg, hdrs)
648 class HTTPDefaultErrorHandler(BaseHandler):
649 def http_error_default(self, req, fp, code, msg, hdrs):
--> 650 raise HTTPError(req.full_url, code, msg, hdrs, fp)
651
652 class HTTPRedirectHandler(BaseHandler):

HTTPError: HTTP Error 403: Forbidden

[qiminchen]

Having the same issue here when trying to download pretrained vgg16 model

[xta0]

Same here on macOS. curl works for me though. Any ideas?

[kawashimaken]

same here
I am trying to use VGG19
File "vgg19.py", line 54, in _step1_load_model self.model = models.vgg19(pretrained=True) File "/opt/anaconda3/lib/python3.7/site-packages/torchvision/models/vgg.py", line 172, in vgg19 return _vgg('vgg19', 'E', False, pretrained, progress, **kwargs) File "/opt/anaconda3/lib/python3.7/site-packages/torchvision/models/vgg.py", line 93, in _vgg progress=progress) File "/opt/anaconda3/lib/python3.7/site-packages/torch/hub.py", line 492, in load_state_dict_from_url download_url_to_file(url, cached_file, hash_prefix, progress=progress) File "/opt/anaconda3/lib/python3.7/site-packages/torch/hub.py", line 391, in download_url_to_file u = urlopen(url) File "/opt/anaconda3/lib/python3.7/urllib/request.py", line 222, in urlopen return opener.open(url, data, timeout) File "/opt/anaconda3/lib/python3.7/urllib/request.py", line 531, in open response = meth(req, response) File "/opt/anaconda3/lib/python3.7/urllib/request.py", line 641, in http_response 'http', request, response, code, msg, hdrs) File "/opt/anaconda3/lib/python3.7/urllib/request.py", line 569, in error return self._call_chain(*args) File "/opt/anaconda3/lib/python3.7/urllib/request.py", line 503, in _call_chain result = func(*args) File "/opt/anaconda3/lib/python3.7/urllib/request.py", line 649, in http_error_default raise HTTPError(req.full_url, code, msg, hdrs, fp) urllib.error.HTTPError: HTTP Error 403: Forbidden

[reiinakano]

Also happens in torch hub.

This is the official notebook example which fails with the same error
https://colab.research.google.com/github/pytorch/pytorch.github.io/blob/master/assets/hub/pytorch_vision_googlenet.ipynb#scrollTo=QjN0iFB8vAjw

[vdantu]

+1 .. I see the same issue with 0.4.2

[MrGrayCode]

https://github.com/pytorch/vision/blob/0156d58ec867590b1c78fe1bc834c7da9afdf46a/torchvision/models/resnet.py#L11-L21
A temporary workaround would be to update the model_urls prefixes from https://download.pytorch.org/models/ to https://s3.amazonaws.com/pytorch/models/

[jonasrauber]

For me, this only fails on the CI server (GitHub actions); locally the download works
EDIT: it works from some machines and fails on others (using Python / torchvision).
(wget always works)

[lopuhin]

Looks like it started working again.

[danielmccluskey]

Working again for me as well now.

[kawashimaken]

It is working now.

[lopuhin]

And now it's failing with URLError: <urlopen error [SSL: CERTIFICATE_VERIFY_FAILED] certificate verify failed (_ssl.c:852)> (again multiple locations, Python 3.6.9)

[uloma07]

It's started failing again now:

Downloading: "https://download.pytorch.org/models/vgg19-dcbb9e9d.pth" to /root/.cache/torch/checkpoints/vgg19-dcbb9e9d.pth

SSLError Traceback (most recent call last)
/usr/lib/python3.6/urllib/request.py in do_open(self, http_class, req, **http_conn_args)
1317 h.request(req.get_method(), req.selector, req.data, headers,
-> 1318 encode_chunked=req.has_header('Transfer-encoding'))
1319 except OSError as err: # timeout error

20 frames
SSLError: [SSL: CERTIFICATE_VERIFY_FAILED] certificate verify failed (_ssl.c:852)

During handling of the above exception, another exception occurred:

URLError Traceback (most recent call last)
/usr/lib/python3.6/urllib/request.py in do_open(self, http_class, req, **http_conn_args)
1318 encode_chunked=req.has_header('Transfer-encoding'))
1319 except OSError as err: # timeout error
-> 1320 raise URLError(err)
1321 r = h.getresponse()
1322 except:

URLError:

[ajuric]

If you try to open the link for some model (eg. for resenet18 https://download.pytorch.org/models/resnet18-5c106cde.pth) in the browser, you will see an error with "https certificate expired" ...

Please fix this ...

[albanD]

As a temporary workaround, you can use http:// instead of https:// to download the model.
EDIT: should be fixed now. Can you confirm it works fine now?

[uloma07]

As a temporary workaround, you can use http:// instead of https:// to download the model.
EDIT: should be fixed now. Can you confirm it works fine now?

Works for me.

[soumith]

The SSL Certificate issue is fixed now. The Cert was expired for ~4 hours, and we root-caused the auto-renewal problem, it wont happen again.

[savourylie]

I'm getting something similar when trying to download torchvision.models.segmentation.deeplabv3_resnet50():

<urlopen error [SSL: CERTIFICATE_VERIFY_FAILED] certificate verify failed: unable to get local issuer certificate (_ssl.c:1076)>

[fmassa]

@savourylie is this still an issue?

[ibrahimjelliti]

adding hub/ in the path after /.cache/torch/ worked for me

[thesyntaxinator]

Still doesn't work for me

[izdeby]

Reopening as its reported again.

[bdthombre]

Downloading: "https://download.pytorch.org/models/resnet18-5c106cde.pth" to C:\Users\bdtho/.cache\torch\hub\checkpoints\resnet18-5c106cde.pth

SSLCertVerificationError Traceback (most recent call last)
~.conda\envs\tensorflow_g\lib\urllib\request.py in do_open(self, http_class, req, **http_conn_args)
1349 try:
-> 1350 h.request(req.get_method(), req.selector, req.data, headers,
1351 encode_chunked=req.has_header('Transfer-encoding'))

~.conda\envs\tensorflow_g\lib\http\client.py in request(self, method, url, body, headers, encode_chunked)
1254 """Send a complete request to the server."""
-> 1255 self._send_request(method, url, body, headers, encode_chunked)
1256

~.conda\envs\tensorflow_g\lib\http\client.py in _send_request(self, method, url, body, headers, encode_chunked)
1300 body = _encode(body, 'body')
-> 1301 self.endheaders(body, encode_chunked=encode_chunked)
1302

~.conda\envs\tensorflow_g\lib\http\client.py in endheaders(self, message_body, encode_chunked)
1249 raise CannotSendHeader()
-> 1250 self._send_output(message_body, encode_chunked=encode_chunked)
1251

~.conda\envs\tensorflow_g\lib\http\client.py in _send_output(self, message_body, encode_chunked)
1009 del self._buffer[:]
-> 1010 self.send(msg)
1011

~.conda\envs\tensorflow_g\lib\http\client.py in send(self, data)
949 if self.auto_open:
--> 950 self.connect()
951 else:

~.conda\envs\tensorflow_g\lib\http\client.py in connect(self)
1423
-> 1424 self.sock = self._context.wrap_socket(self.sock,
1425 server_hostname=server_hostname)

~.conda\envs\tensorflow_g\lib\ssl.py in wrap_socket(self, sock, server_side, do_handshake_on_connect, suppress_ragged_eofs, server_hostname, session)
499 # ctx._wrap_socket()
--> 500 return self.sslsocket_class._create(
501 sock=sock,

~.conda\envs\tensorflow_g\lib\ssl.py in _create(cls, sock, server_side, do_handshake_on_connect, suppress_ragged_eofs, server_hostname, context, session)
1039 raise ValueError("do_handshake_on_connect should not be specified for non-blocking sockets")
-> 1040 self.do_handshake()
1041 except (OSError, ValueError):

~.conda\envs\tensorflow_g\lib\ssl.py in do_handshake(self, block)
1308 self.settimeout(None)
-> 1309 self._sslobj.do_handshake()
1310 finally:

SSLCertVerificationError: [SSL: CERTIFICATE_VERIFY_FAILED] certificate verify failed: unable to get issuer certificate (_ssl.c:1123)

During handling of the above exception, another exception occurred:

URLError Traceback (most recent call last)
in
----> 1 model = torchvision.models.resnet18(pretrained=True)

~.conda\envs\tensorflow_g\lib\site-packages\torchvision\models\resnet.py in resnet18(pretrained, progress, *kwargs)
238 progress (bool): If True, displays a progress bar of the download to stderr
239 """
--> 240 return _resnet('resnet18', BasicBlock, [2, 2, 2, 2], pretrained, progress,
241 *kwargs)
242

~.conda\envs\tensorflow_g\lib\site-packages\torchvision\models\resnet.py in _resnet(arch, block, layers, pretrained, progress, *kwargs)
224 model = ResNet(block, layers, *kwargs)
225 if pretrained:
--> 226 state_dict = load_state_dict_from_url(model_urls[arch],
227 progress=progress)
228 model.load_state_dict(state_dict)

~.conda\envs\tensorflow_g\lib\site-packages\torch\hub.py in load_state_dict_from_url(url, model_dir, map_location, progress, check_hash, file_name)
553 r = HASH_REGEX.search(filename) # r is Optional[Match[str]]
554 hash_prefix = r.group(1) if r else None
--> 555 download_url_to_file(url, cached_file, hash_prefix, progress=progress)
556
557 if _is_legacy_zip_format(cached_file):

~.conda\envs\tensorflow_g\lib\site-packages\torch\hub.py in download_url_to_file(url, dst, hash_prefix, progress)
423 # certificates in older Python
424 req = Request(url, headers={"User-Agent": "torch.hub"})
--> 425 u = urlopen(req)
426 meta = u.info()
427 if hasattr(meta, 'getheaders'):

~.conda\envs\tensorflow_g\lib\urllib\request.py in urlopen(url, data, timeout, cafile, capath, cadefault, context)
220 else:
221 opener = _opener
--> 222 return opener.open(url, data, timeout)
223
224 def install_opener(opener):

~.conda\envs\tensorflow_g\lib\urllib\request.py in open(self, fullurl, data, timeout)
523
524 sys.audit('urllib.Request', req.full_url, req.data, req.headers, req.get_method())
--> 525 response = self._open(req, data)
526
527 # post-process response

~.conda\envs\tensorflow_g\lib\urllib\request.py in _open(self, req, data)
540
541 protocol = req.type
--> 542 result = self._call_chain(self.handle_open, protocol, protocol +
543 '_open', req)
544 if result:

~.conda\envs\tensorflow_g\lib\urllib\request.py in _call_chain(self, chain, kind, meth_name, args)
500 for handler in handlers:
501 func = getattr(handler, meth_name)
--> 502 result = func(args)
503 if result is not None:
504 return result

~.conda\envs\tensorflow_g\lib\urllib\request.py in https_open(self, req)
1391
1392 def https_open(self, req):
-> 1393 return self.do_open(http.client.HTTPSConnection, req,
1394 context=self._context, check_hostname=self._check_hostname)
1395

~.conda\envs\tensorflow_g\lib\urllib\request.py in do_open(self, http_class, req, **http_conn_args)
1351 encode_chunked=req.has_header('Transfer-encoding'))
1352 except OSError as err: # timeout error
-> 1353 raise URLError(err)
1354 r = h.getresponse()
1355 except:

URLError:

[Saninsusanin]

I have tried to download pre-trained vgg16 on local machine(MacOS 10.15) and got this stack:

Downloading:` "https://download.pytorch.org/models/vgg19-dcbb9e9d.pth" to /Users/root/.cache/torch/hub/checkpoints/vgg19-dcbb9e9d.pth
…
ssl.SSLCertVerificationError: [SSL: CERTIFICATE_VERIFY_FAILED] certificate verify failed: unable to get local issuer certificate (_ssl.c:1076)

And, by the way, I can download it from colab

[datumbox]

This might be related to S3 issues in different regions. This could explain why the models can be accessed from some regions while not from others. I tried both URLs and I can download the files just fine...