Sagemaker Studio API to get original s3 dataset path of an AutoMLJob
0
Is there any API of Sagemaker Studio that can help us get the complete path of the original dataset that was imported from s3 in Canvas for model training, in an autoMLJob?
I have imported a dataset from my own s3 bucket in Canvas and trained a model using that dataset. When the training is triggered, Canvas saves the copy of that original dataset to a new path in s3, trains the model on that dataset and saves the path of that dataset in the autoMLJob of that model in Sagemaker Studio under the tag InputDataConfig. Is there any way that we can get the s3 path of the original dataset too from where we have imported the dataset, through any API of Sagemaker Studio or through the describeAutoMLJob API?
Are these answers helpful? Upvote the correct answer to help the community benefit from your knowledge.
0
Hi Aamna,
You can use the describeAutoMLJob API to retrieve the original datasource that you have input for the job. It falls under the tag [InputDataConfig][Datasource][S3Uri].
Here is a link for your reference to the API's response syntax.
I have tried the describeAutoMLJob API but under the tag [InputDataConfig][Datasource][S3Uri] I am getting the path of the copy of dataset that Canvas have created and saved for the training of model, I need the path/information of the actual dataset from which the copy is created. Do you have any idea about that?
Unfortunately this information is not captured currently. You may name the imported dataset in a way that helps you associate with the original dataset from S3 easily.
AWS OFFICIALUpdated a year ago
Hi Junkai, Thankyou for your response.
I have tried the describeAutoMLJob API but under the tag [InputDataConfig][Datasource][S3Uri] I am getting the path of the copy of dataset that Canvas have created and saved for the training of model, I need the path/information of the actual dataset from which the copy is created. Do you have any idea about that?
Hi Aamna,
Unfortunately this information is not captured currently. You may name the imported dataset in a way that helps you associate with the original dataset from S3 easily.
Thankyou for your response Junkai. Much appreciated.