I'm just trying out a tutorial (https://cloud.google.com/ml-engine/docs/tensorflow/flowers-tutorial) to get familiar with model training & scoring for classification problems using ML compute engine. But soon after declaring variable vals for path, region, bucket name, etc., as i reach the step to run pre-processing script. The cloud shell automatically closes or shuts down for some reasons, not sure. But soon as i reopen it to proceed with remaining steps, vals of all previously declared variable is NULL again. So is there some capping/problem with GCP cloudshell, or i'm missing setting some parameter? Any help, i'd really appreciate! I'm a newbie to cloud training. Thanks.
Asked
Active
Viewed 670 times
0
-
Use a session program such as `screen`, `tmux`, etc. This will preserve your sessions so that you can reconnect to a previous session. However for long running development / jobs, setup the development environment on your desktop. – John Hanley Feb 23 '19 at 00:26
-
following same tutorial, soon as i execute: python trainer/preprocess.py --input_dict "$DICT_FILE" --input_path "gs://cloud-ml-data/img/flower_photos/eval_set.csv" --output_path "${GCS_PATH}/preproc/eval" --cloud on bash locally, the window vanishes in the same manner as it was happening with browser based cloudshell. I'm still not getting why it's happening? Working with Cloud SDK is advised for further development. Please help me reason this behaviour of cloudhshell/bash and mitigate issues if any. or perhaps any alternate how i can experiment with Gcloud ML engine tutorials.? Thanks – Prakash Mar 06 '19 at 07:10