Read Error while loading vgg16

I am loading a pretrained model of vgg16:
vgg16=torchvision.models.vgg16(pretrained=True)
I am getting the following error:

---------------------------------------------------------------------------
ReadError                                 Traceback (most recent call last)
<ipython-input-21-48bdd58aa112> in <module>()
----> 1 vgg16=torchvision.models.vgg16(pretrained=True)

/home/sarthak/anaconda2/lib/python2.7/site-packages/torchvision/models/vgg.pyc in vgg16(pretrained, **kwargs)
    122     model = VGG(make_layers(cfg['D']), **kwargs)
    123     if pretrained:
--> 124         model.load_state_dict(model_zoo.load_url(model_urls['vgg16']))
    125     return model
    126 

/home/sarthak/anaconda2/lib/python2.7/site-packages/torch/utils/model_zoo.pyc in load_url(url, model_dir)
     55         hash_prefix = HASH_REGEX.search(filename).group(1)
     56         _download_url_to_file(url, cached_file, hash_prefix)
---> 57     return torch.load(cached_file)
     58 
     59 

/home/sarthak/anaconda2/lib/python2.7/site-packages/torch/serialization.pyc in load(f, map_location, pickle_module)
    246         f = open(f, 'rb')
    247     try:
--> 248         return _load(f, map_location, pickle_module)
    249     finally:
    250         if new_fd:

/home/sarthak/anaconda2/lib/python2.7/site-packages/torch/serialization.pyc in _load(f, map_location, pickle_module)
    312         return deserialized_objects[int(saved_id)]
    313 
--> 314     with closing(tarfile.open(fileobj=f, mode='r:', format=tarfile.PAX_FORMAT)) as tar, \
    315          mkdtemp() as tmpdir:
    316 

/home/sarthak/anaconda2/lib/python2.7/tarfile.pyc in open(cls, name, mode, fileobj, bufsize, **kwargs)
   1691             else:
   1692                 raise CompressionError("unknown compression type %r" % comptype)
-> 1693             return func(name, filemode, fileobj, **kwargs)
   1694 
   1695         elif "|" in mode:

/home/sarthak/anaconda2/lib/python2.7/tarfile.pyc in taropen(cls, name, mode, fileobj, **kwargs)
   1721         if mode not in ("r", "a", "w"):
   1722             raise ValueError("mode must be 'r', 'a' or 'w'")
-> 1723         return cls(name, mode, fileobj, **kwargs)
   1724 
   1725     @classmethod

/home/sarthak/anaconda2/lib/python2.7/tarfile.pyc in __init__(self, name, mode, fileobj, format, tarinfo, dereference, ignore_zeros, encoding, errors, pax_headers, debug, errorlevel)
   1585             if self.mode == "r":
   1586                 self.firstmember = None
-> 1587                 self.firstmember = self.next()
   1588 
   1589             if self.mode == "a":

/home/sarthak/anaconda2/lib/python2.7/tarfile.pyc in next(self)
   2368                     continue
   2369                 elif self.offset == 0:
-> 2370                     raise ReadError(str(e))
   2371             except EmptyHeaderError:
   2372                 if self.offset == 0:

ReadError: invalid header

Either:

  1. you are not on the latest pytorch 0.1.11
    or
  2. The download didn’t finish and the file is corrupted. Clear $HOME/.torch

I deleted the file and then downloaded it again but still the same error
I get the same error with vgg19 but not with resnet50 and alexnet
My pytorch version is 0.1.7. How do I get the latest version.
I tried conda update pytorch but it says up to date.

Thanks for the help. I updated the pytorch version and now it works