Search training plan offerings - Amazon SageMaker AI
Required ParametersOptional Parameters

Search training plan offerings

To create a training plan, start by calling the SearchTrainingPlanOfferings API operation, passing your plan requirements (such as instance type, count, and desired time window) as input parameters. Training plans are specific to their target resource. Ensure that you specify which target resource the plan will be used for (training-job or hyperpod-cluster). The API returns a list of available offerings that match your requirements. If no suitable offerings are found, you may need to adjust your requirements and search again.

This API call retrieves the training plan offerings that best meet your capacity needs. Each TrainingPlanOffering returned in the response is identified by a unique offering ID. The first offering in the list represents the best match for your requirements. If no suitable training plan is available within your specified dates, the list is empty. Adjust your search criteria and look for a new set of offerings.

Important

You can use SageMaker training plans to reserve instances with the following reservation duration and instance quantity options.

  • Reservation durations are available in 1-day increments from 1 to 182 days.

  • The reservation instance quantity options are 1, 2, 4, 8, 16, 32 or 64 instances.

To learn about the list of available instances supported by SageMaker training plans, see Supported instance types and AWS Regions.

The following example uses an AWS CLI command to request training plan offerings with a specified instance type, count, and time information.

# List training plan offerings with instance type, instance count, duration in hours, start time after, and end time before.
aws sagemaker search-training-plan-offerings \
--target-resources "training-job" \
--instance-type "ml.p5.48xlarge" \
--instance-count 4 \
--duration-hours 96 \
--start-time-after "1727838000" \
--end-time-before "1729709600"

This JSON document is a sample response from the SageMaker training plans API. The response provides information about a single available training plan offering that matches the specified capacity requirements.

{
    "TrainingPlanOfferings": [
        {
            "CurrencyCode": "USD",
            "DurationHours": 96,
            "DurationMinutes": 0,
            "RequestedStartTimeAfter": "2024-09-27T18:00:00-07:00",   
            "RequestedEndTimeBefore": "2024-11-23T17:00:00-07:00",
            "ReservedCapacityOfferings": [
                {  
                    "AvailabilityZone": "us-east-1f",
                    "DurationHours": 96,
                    "EndTime": "2024-10-02T04:30:00-07:00",
                    "InstanceType": "ml.p5.48xlarge",
                    "InstanceCount": 4,
                    "StartTime": "2024-09-28T04:30:00-07:00",                   
                }
            ],
            "TargetResources": "training-job",
            "TrainingPlanOfferingId": "tpo-SHA-256-hash-value",
            "UpfrontFee": "xxxx.xx",
        }
    ]
}

The following sections define the mandatory and optional input request parameters for the SearchTrainingPlanOfferings API operation.

Required parameters

When calling the SearchTrainingPlanOfferings API to list training plan offerings that meet your requirements, you must provide the following values:

  • TargetResources: The target resources (training-job or hyperpod-cluster) for which the plan will be used. The default value is training-job. Training plans are specific to their target resource.

    • A training plan designed for SageMaker training jobs can only be used to schedule and run training jobs.

    • A training plan for HyperPod clusters can be used exclusively to provide compute resources to a cluster's instance group.

  • InstanceType: The type of instance to provision. The InstanceType must be of a supported type.

    To learn about the list of available instances supported by SageMaker training plans, see Supported instance types and AWS Regions.

  • InstanceCount: The number of instances to provision. If the number of instances is greater than 1, it should be a power of 2.

Optional parameters

The following sections provide details on some optional parameters that you can pass to your SearchTrainingPlanOfferings API request.

  • DurationHour: The total duration of your requested plan in hours. The DurationHour is rounded up to the nearest multiple of 24.

  • StartTimeAfter: Specify the requested start time of the plan. The StartTimeAfter should be a timestamp or an ISO 8601 date/time value in the future.

  • EndTimeBefore: Specify the requested end time of the plan in a timestamp or an ISO 8601 date/time format. The EndTimeBefore should be at least 24 hours after the start time .