Getting error when trying to load data using keras.utils.get_file()

Question

I am getting an error when trying to load a dataset using TensorFlow Keras. Here is the code:

dataset_url = "https://storage.googleapis.com/sample_org/sample_file.zip"
data_dir = tf.keras.utils.get_file(origin=dataset_url, 
                                   fname='sample_file', 
                                   extract=True)
data_dir = pathlib.Path(data_dir)

I have changed the URL to 'sample_file.zip' for security reasons.

This is the error that I am getting:

---------------------------------------------------------------------------
FileExistsError                           Traceback (most recent call last)
<ipython-input-12-1cdc186b0389> in <module>()
      2 data_dir = tf.keras.utils.get_file(origin=dataset_url, 
      3                                    fname='sample_file',
----> 4                                    extract=True)
      5 data_dir = pathlib.Path(data_dir)

3 frames
/usr/lib/python3.6/zipfile.py in _extract_member(self, member, targetpath, pwd)
   1572         if member.is_dir():
   1573             if not os.path.isdir(targetpath):
-> 1574                 os.mkdir(targetpath)
   1575             return targetpath
   1576 

FileExistsError: [Errno 17] File exists: '/root/.keras/datasets/sample_file'

What is causing this error? How can I fix it?

This is the first block of code run after imports so I don't know why 'FileExistsError' happens.

I have tried changing the name of the file.

I have checked TensorFlow documentation and it uses code like this:

dataset_url = "https://storage.googleapis.com/download.tensorflow.org/example_images/flower_photos.tgz"
data_dir = tf.keras.utils.get_file(origin=dataset_url, 
                                   fname='flower_photos', 
                                   untar=True)
data_dir = pathlib.Path(data_dir)

I have run this above code and it is working. But I cannot figure out why the same code is showing error for my data. Please advice.

Answer 1

I think you have a file (without any extension) called /root/.keras/datasets/sample_file , which is why there is an error. Try giving it a different name fname='my_very_unique_sample_file_name_that_does_not_exist_yet'