Forbidden S3 bucket in the example: amazon-sagemaker-examples/introduction_to_applying_machine_learning/gluon_recommender_system/gluon_recommender_system.ipynb about amazon-sagemaker-examples HOT 4 CLOSED

A temporary solution I found (See this discussion) is running the s3 cp command with --no-sign-request parameter as shown below:
aws s3 --no-sign-request cp s3://amazon-reviews-pds/tsv/amazon_reviews_us_Digital_Video_Download_v1_00.tsv.gz /tmp/recsys/

from amazon-sagemaker-examples.

After fixing the S3 issue, I tested the rest of the code but the kernel (Both conda_mxnet_p27 and conda_mxnet_p36) keeps dying at Let's initialize network weights and set our optimization parameters section. See the attached screenshot. Could you advise what could be a cause of this problem?

from amazon-sagemaker-examples.

Hi @pilhokim . Could you confirm that you're running in a GPU instance? The MXNet context is currently set at mx.gpu() which will fail if you're running in a CPU instance. Thanks.

from amazon-sagemaker-examples.

You are right, @djarpin! Thanks a lot for your advice. I forgot the fact that this code runs on the local machine, not deploying training instances. I tested ctx = mx.cpu() which works fine with my ml.m4x.xlarge SageMaker instance. I also created a ml.p2.xlarge SageMaker instance and set ctx = mx.gpu. Both work fine. This example, however, looks taking a long time to train. Thank you again.

from amazon-sagemaker-examples.