549957 (2) [Avatar] Offline
#1
imdb <- dataset_imdb(num_words = 10000)

I get the following error message:
" Error in py_call_impl(callable, dots$args, dots$keywords) :
Exception: URL fetch failure on https://s3.amazonaws.com/text-datasets/imdb.npz: None -- unknown url type: https"

I am able to successfully load the mnist data set with 'data <- dataset_mnist()'

Running Ubuntu 16.04, Rstudio and R keras with GPU support. On the python end, running python3.6 (anaconda).

Is anyone else having a similar problem? Solution?
549957 (2) [Avatar] Offline
#2
One additional point:
I am able to successfully load and use the imdb dataset in python (python3.6) using keras with tensorflow backend.
So I don't think my problem is with the python version, but perhaps rather with the R keras interface to python.
Thank you
553979 (1) [Avatar] Offline
#3
I had the same problem. Can't figure out how to get it working but a workaround is to download the dataset

wget https://s3.amazonaws.com/img-datasets/mnist.npz

and then load the dataset with
mnist <- dataset_mnist('/path/to/mnist.npz')