Error with time series chart of trial components under experiments

0

I'm using the functionality of sagemaker.experiments, where a run object is defined for tracking a job. For logging of metrics, I'm using the log_metric() method of the run object, with name, value and step (epoch).

Under the sagemaker studio UI, I can see the experiment and the relevant runs, I also can see aggregations of the metrics (such as min, max, average, count, etc.), but when I try to plot a chart of a metric over epoch (or over time), there is an error with this message:

An error has occurred while retrieving chart data.

Particularly, I've run this example notebook, which uses local mode for training a model and logging metrics. https://sagemaker-examples.readthedocs.io/en/latest/sagemaker-experiments/local_experiment_tracking/pytorch_experiment.html

I ran this notebook as is!

When running this notebook under my sagemaker domain, I can see that the experiment and runs are logged with correct parameters and artifacts, but still the metrics tab only shows the summary of metrics and not the full data. I can;t plot charts of metrics against epochs, as before.

Any idea what is going on here? Maybe there is something else needed?

Yair
asked a year ago229 views
2 Answers
1

Hello,

Thank you for using AWS SageMaker.

From the correspondence, I understand you are encountering issues while plotting the charts after running the example mentioned for the SageMaker experiment example shared above.

While executing the same example under my test account, I was able to plot the chart without any error and without any extra commands added to the actual script. To get more insights on the issue that has occurred at your end, we encourage you to reach out to AWS Support by opening case so that the engineers can assist your case and help you overcome the issue. To open a support case with AWS using the link: https://console.aws.amazon.com/support/home?#/case/create

AWS
SUPPORT ENGINEER
answered a year ago
0

Thanks for the response.

I've opened a support case, and after some trials, I figured out that all I was needed is to shut down the Studio and re-open it.

This resolved the problem

Yair
answered a year ago

You are not logged in. Log in to post an answer.

A good answer clearly answers the question and provides constructive feedback and encourages professional growth in the question asker.

Guidelines for Answering Questions