Using custom containers with AI Platform Training Reviews
4209 reviews
gianpiero p. · Reviewed over 5 years ago
error.. error .. error
Alper K. · Reviewed over 5 years ago
Or I. · Reviewed over 5 years ago
Raghav K. · Reviewed over 5 years ago
Radomir P. · Reviewed over 5 years ago
Encountered multiple errors First you need to do cd home/jupyter before git clone command Second the schema in bigquery gave an error, had to change the $SCHEMA variable and define it in one line Third got an error on gcloud builds submit: ERROR: After October 2020 you may experience errors when installing or updating packages. This is because pip will change the way that it resolves dependency conflicts. We recommend you use --use-feature=2020-resolver to test your packages with the new resolver before it becomes the default. visions 0.6.4 requires pandas>=0.25.3, but you'll have pandas 0.24.2 which is incompatible. pandas-profiling 2.8.0 requires pandas!=1.0.0,!=1.0.1,!=1.0.2,>=0.25.3, but you'll have pandas 0.24.2 which is incompatible. pandas-profiling 2.8.0 requires visions[type_image_path]==0.4.4, but you'll have visions 0.6.4 which is incompatible. Fourth got an error when submitting the training job: ERROR: (gcloud.ai-platform.jobs.submit.training) argument --job-dir: Must be of form gs://bucket/object
Fran M. · Reviewed over 5 years ago
Deepak S. · Reviewed over 5 years ago
There are certain errors in the instructions, also the end checks are not working properly.
Amita K. · Reviewed over 5 years ago
I found this lab buggy, lengthy, and with wrong explanation. Even if I was able to complete the task, the score did not update.
Médéric H. · Reviewed over 5 years ago
Some issue: - some step for CloudShell are not needed - why do we create a Kubeflow pipeline if none of the steps are run there (Notebook ....) ? - some issue for the creation of the BQ table because of extra space for each line - broken grading for hp step
Dr. F. · Reviewed over 5 years ago
Galip H. · Reviewed over 5 years ago
Irfan J. · Reviewed over 5 years ago
Irfan J. · Reviewed over 5 years ago
Some issue: - some step for CloudShell are not needed - why do we create a Kubeflow pipeline if none of the steps are run there (Notebook ....) ? - some issue for the creation of the BQ table because of extra space for each line - broken grading for hp step
Dr. F. · Reviewed over 5 years ago
Lab is not complete!
Arka A. · Reviewed over 5 years ago
It has a bug. But service support is working on it
Ivan N. · Reviewed over 5 years ago
Not able to check "Prepare the hyperparameter tuning application". step
Surachart O. · Reviewed over 5 years ago
Jinho S. · Reviewed over 5 years ago
Surachart O. · Reviewed over 5 years ago
One more time!
Stewart A. · Reviewed over 5 years ago
Ran out of time!
Stewart A. · Reviewed over 5 years ago
Ajay H. · Reviewed over 5 years ago
The notebook directory needs to be the shell's wd prior to the github download. Also the model and version create step logic does not take into account the "Using endpoint [https://ml.googleapis.com/]" command output.
Stewart A. · Reviewed over 5 years ago
The exercise lab notebook was inaccessible.
Stewart A. · Reviewed over 5 years ago
not finish yet. got some error on notebook. BigQuery error in extract operation: Error processing job 'qwiklabs- gcp-01-f90870621d28:bqjob_r2125cf07fbc1c587_00000175899dc6a4_1': Access Denied: BigQuery BigQuery: Permission denied while writing data.
Surachart O. · Reviewed over 5 years ago
We do not ensure the published reviews originate from consumers who have purchased or used the products. Reviews are not verified by Google.