AWS sagemaker documentation change
Summary
Updated algorithm/DLC listings and registry paths for US Gov East region, added new frameworks (DJL DeepSpeed, Inferentia variants, Neo variants, Spark), updated SDK examples, and expanded version support
Security assessment
Changes consist of routine documentation updates for framework support and container registry paths. No security advisories, vulnerability fixes, or security configuration guidance were added/modified. The additions focus on expanding supported ML frameworks and versions rather than addressing security concerns.
Diff
diff --git a/sagemaker/latest/dg-ecr-paths/ecr-us-gov-east-1.md b/sagemaker/latest/dg-ecr-paths/ecr-us-gov-east-1.md index 958523fbe..c76d1022e 100644 --- a//sagemaker/latest/dg-ecr-paths/ecr-us-gov-east-1.md +++ b//sagemaker/latest/dg-ecr-paths/ecr-us-gov-east-1.md @@ -5 +5 @@ -AutoGluonBlazingTextDeepAR ForecastingFactorization MachinesHugging FaceIP InsightsImage classificationK-MeansKNNLinear LearnerMXNetNTMObject DetectionObject2VecPCAPyTorchPyTorch NeuronRandom Cut ForestScikit-learnSemantic SegmentationSeq2SeqTensorflowXGBoost +AutoGluonBlazingTextDJL DeepSpeedDeepAR ForecastingFactorization MachinesHugging FaceIP InsightsImage classificationInferentia MXNetInferentia PyTorchK-MeansKNNLinear LearnerMXNetNTMNeo Image ClassificationNeo MXNetNeo PyTorchNeo TensorflowNeo XGBoostObject DetectionObject2VecPCAPyTorchRandom Cut ForestScikit-learnSemantic SegmentationSeq2SeqSparkTensorflowTensorflow InferentiaXGBoost @@ -9 +9 @@ AutoGluonBlazingTextDeepAR ForecastingFactorization MachinesHugging FaceIP Insig -The following topics list parameters for each of the algorithms and deep learning containers in this region provided by Amazon SageMaker AI. +The following topics list parameters for each of the algorithms and deep learning containers that are provided by Amazon SageMaker AI in this AWS Region. @@ -16,0 +17,2 @@ The following topics list parameters for each of the algorithms and deep learnin + * DJL DeepSpeed (algorithm) + @@ -26,0 +29,4 @@ The following topics list parameters for each of the algorithms and deep learnin + * Inferentia MXNet (DLC) + + * Inferentia PyTorch (DLC) + @@ -36,0 +43,10 @@ The following topics list parameters for each of the algorithms and deep learnin + * Neo Image Classification (algorithm) + + * Neo MXNet (DLC) + + * Neo PyTorch (DLC) + + * Neo Tensorflow (DLC) + + * Neo XGBoost (algorithm) + @@ -45,2 +60,0 @@ The following topics list parameters for each of the algorithms and deep learnin - * PyTorch Neuron (DLC) - @@ -54,0 +69,2 @@ The following topics list parameters for each of the algorithms and deep learnin + * Spark (algorithm) + @@ -56,0 +73,2 @@ The following topics list parameters for each of the algorithms and deep learnin + * Tensorflow Inferentia (DLC) + @@ -64 +82 @@ The following topics list parameters for each of the algorithms and deep learnin -SageMaker AI Python SDK example to retrieve registry path. +The following SageMaker AI Python SDK example shows how to retrieve a specific registry path. @@ -71,0 +90,12 @@ Registry path | Version | Job types (image scope) +446045086412.dkr.ecr.us-gov-east-1.amazonaws.com/autogluon-training:`<tag>` | 1.3.0 | training +446045086412.dkr.ecr.us-gov-east-1.amazonaws.com/autogluon-inference:`<tag>` | 1.3.0 | inference +446045086412.dkr.ecr.us-gov-east-1.amazonaws.com/autogluon-training:`<tag>` | 1.2.0 | training +446045086412.dkr.ecr.us-gov-east-1.amazonaws.com/autogluon-inference:`<tag>` | 1.2.0 | inference +446045086412.dkr.ecr.us-gov-east-1.amazonaws.com/autogluon-training:`<tag>` | 1.1.1 | training +446045086412.dkr.ecr.us-gov-east-1.amazonaws.com/autogluon-inference:`<tag>` | 1.1.1 | inference +446045086412.dkr.ecr.us-gov-east-1.amazonaws.com/autogluon-training:`<tag>` | 1.1.0 | training +446045086412.dkr.ecr.us-gov-east-1.amazonaws.com/autogluon-inference:`<tag>` | 1.1.0 | inference +446045086412.dkr.ecr.us-gov-east-1.amazonaws.com/autogluon-training:`<tag>` | 1.0.0 | training +446045086412.dkr.ecr.us-gov-east-1.amazonaws.com/autogluon-inference:`<tag>` | 1.0.0 | inference +446045086412.dkr.ecr.us-gov-east-1.amazonaws.com/autogluon-training:`<tag>` | 0.8.2 | training +446045086412.dkr.ecr.us-gov-east-1.amazonaws.com/autogluon-inference:`<tag>` | 0.8.2 | inference @@ -93 +123 @@ Registry path | Version | Job types (image scope) -SageMaker AI Python SDK example to retrieve registry path. +The following SageMaker AI Python SDK example shows how to retrieve a specific registry path. @@ -102,0 +133,20 @@ Registry path | Version | Job types (image scope) +## DJL DeepSpeed (algorithm) + +The following SageMaker AI Python SDK example shows how to retrieve a specific registry path. + + + from sagemaker import image_uris + image_uris.retrieve(framework='djl-deepspeed', region='us-west-2',py_version='py3',image_scope='inference') + +Registry path | Version | Job types (image scope) +---|---|--- +446045086412.dkr.ecr.us-gov-east-1.amazonaws.com/djl-inference:0.27.0-deepspeed0.12.6-cu121-`<tag>` | 0.27.0 | inference +446045086412.dkr.ecr.us-gov-east-1.amazonaws.com/djl-inference:0.26.0-deepspeed0.12.6-cu121-`<tag>` | 0.26.0 | inference +446045086412.dkr.ecr.us-gov-east-1.amazonaws.com/djl-inference:0.25.0-deepspeed0.11.0-cu118-`<tag>` | 0.25.0 | inference +446045086412.dkr.ecr.us-gov-east-1.amazonaws.com/djl-inference:0.24.0-deepspeed0.10.0-cu118-`<tag>` | 0.24.0 | inference +446045086412.dkr.ecr.us-gov-east-1.amazonaws.com/djl-inference:0.23.0-deepspeed0.9.5-cu118-`<tag>` | 0.23.0 | inference +446045086412.dkr.ecr.us-gov-east-1.amazonaws.com/djl-inference:0.22.1-deepspeed0.9.2-cu118-`<tag>` | 0.22.1 | inference +446045086412.dkr.ecr.us-gov-east-1.amazonaws.com/djl-inference:0.21.0-deepspeed0.8.3-cu117-`<tag>` | 0.21.0 | inference +446045086412.dkr.ecr.us-gov-east-1.amazonaws.com/djl-inference:0.20.0-deepspeed0.7.5-cu116-`<tag>` | 0.20.0 | inference +446045086412.dkr.ecr.us-gov-east-1.amazonaws.com/djl-inference:0.19.0-deepspeed0.7.3-cu113-`<tag>` | 0.19.0 | inference + @@ -105 +155 @@ Registry path | Version | Job types (image scope) -SageMaker AI Python SDK example to retrieve registry path. +The following SageMaker AI Python SDK example shows how to retrieve a specific registry path. @@ -117 +167 @@ Registry path | Version | Job types (image scope) -SageMaker AI Python SDK example to retrieve registry path. +The following SageMaker AI Python SDK example shows how to retrieve a specific registry path. @@ -129 +179 @@ Registry path | Version | Job types (image scope) -SageMaker AI Python SDK example to retrieve registry path. +The following SageMaker AI Python SDK example shows how to retrieve a specific registry path. @@ -136,0 +187,9 @@ Registry path | Version | Job types (image scope) +446045086412.dkr.ecr.us-gov-east-1.amazonaws.com/huggingface-pytorch-training:`<tag>` | 4.49.0 | training +446045086412.dkr.ecr.us-gov-east-1.amazonaws.com/huggingface-pytorch-inference:`<tag>` | 4.49.0 | inference +446045086412.dkr.ecr.us-gov-east-1.amazonaws.com/huggingface-pytorch-training:`<tag>` | 4.48.0 | training +446045086412.dkr.ecr.us-gov-east-1.amazonaws.com/huggingface-pytorch-inference:`<tag>` | 4.48.0 | inference +446045086412.dkr.ecr.us-gov-east-1.amazonaws.com/huggingface-pytorch-training:`<tag>` | 4.46.1 | training +446045086412.dkr.ecr.us-gov-east-1.amazonaws.com/huggingface-pytorch-inference:`<tag>` | 4.37.0 | inference +446045086412.dkr.ecr.us-gov-east-1.amazonaws.com/huggingface-pytorch-training:`<tag>` | 4.36.0 | training +446045086412.dkr.ecr.us-gov-east-1.amazonaws.com/huggingface-pytorch-training:`<tag>` | 4.28.1 | training +446045086412.dkr.ecr.us-gov-east-1.amazonaws.com/huggingface-pytorch-inference:`<tag>` | 4.28.1 | inference @@ -173 +232 @@ Registry path | Version | Job types (image scope) -SageMaker AI Python SDK example to retrieve registry path. +The following SageMaker AI Python SDK example shows how to retrieve a specific registry path. @@ -185 +244 @@ Registry path | Version | Job types (image scope) -SageMaker AI Python SDK example to retrieve registry path. +The following SageMaker AI Python SDK example shows how to retrieve a specific registry path. @@ -194,0 +254,27 @@ Registry path | Version | Job types (image scope) +## Inferentia MXNet (DLC) + +The following SageMaker AI Python SDK example shows how to retrieve a specific registry path. + + + from sagemaker import image_uris + image_uris.retrieve(framework='inferentia-mxnet',region='us-gov-east-1',version='1.5.1',instance_type='ml.inf1.6xlarge') + +Registry path | Version | Job types (image scope) | Processor types | Python versions +---|---|---|---|--- +227234621604.dkr.ecr.us-gov-east-1.amazonaws.com/sagemaker-neo-mxnet:`<tag>` | 1.8 | inference | inf | py3 +227234621604.dkr.ecr.us-gov-east-1.amazonaws.com/sagemaker-neo-mxnet:`<tag>` | 1.5.1 | inference | inf | py3 + +## Inferentia PyTorch (DLC) + +The following SageMaker AI Python SDK example shows how to retrieve a specific registry path. + + + from sagemaker import image_uris + image_uris.retrieve(framework='inferentia-pytorch',region='us-gov-east-1',version='1.9',py_version='py3') + +Registry path | Version | Job types (image scope) | Processor types | Python versions +---|---|---|---|--- +227234621604.dkr.ecr.us-gov-east-1.amazonaws.com/sagemaker-neo-pytorch:`<tag>` | 1.9 | inference | inf | py3 +227234621604.dkr.ecr.us-gov-east-1.amazonaws.com/sagemaker-neo-pytorch:`<tag>` | 1.8 | inference | inf | py3 +227234621604.dkr.ecr.us-gov-east-1.amazonaws.com/sagemaker-neo-pytorch:`<tag>` | 1.7 | inference | inf | py3 + @@ -197 +283 @@ Registry path | Version | Job types (image scope) -SageMaker AI Python SDK example to retrieve registry path. +The following SageMaker AI Python SDK example shows how to retrieve a specific registry path. @@ -209 +295 @@ Registry path | Version | Job types (image scope) -SageMaker AI Python SDK example to retrieve registry path. +The following SageMaker AI Python SDK example shows how to retrieve a specific registry path. @@ -221 +307 @@ Registry path | Version | Job types (image scope) -SageMaker AI Python SDK example to retrieve registry path. +The following SageMaker AI Python SDK example shows how to retrieve a specific registry path. @@ -233 +319 @@ Registry path | Version | Job types (image scope) -SageMaker AI Python SDK example to retrieve registry path. +The following SageMaker AI Python SDK example shows how to retrieve a specific registry path. @@ -257 +343 @@ Registry path | Version | Job types (image scope) | Processor types | Python ver -SageMaker AI Python SDK example to retrieve registry path. +The following SageMaker AI Python SDK example shows how to retrieve a specific registry path. @@ -266,0 +353,68 @@ Registry path | Version | Job types (image scope) +## Neo Image Classification (algorithm) + +The following SageMaker AI Python SDK example shows how to retrieve a specific registry path. + + + from sagemaker import image_uris + image_uris.retrieve(framework='image-classification-neo',region='us-gov-east-1') + +Registry path | Version | Job types (image scope) +---|---|--- +227234621604.dkr.ecr.us-gov-east-1.amazonaws.com/image-classification-neo:`<tag>` | latest | inference + +## Neo MXNet (DLC) + +The following SageMaker AI Python SDK example shows how to retrieve a specific registry path. + + + from sagemaker import image_uris + image_uris.retrieve(framework='neo-mxnet',region='us-gov-east-1',version='1.8',py_version='py3',image_scope='inference', instance_type='ml.c5.4xlarge') + +Registry path | Version | Job types (image scope) | Processor types | Python versions +---|---|---|---|--- +227234621604.dkr.ecr.us-gov-east-1.amazonaws.com/sagemaker-inference-mxnet:`<tag>` | 1.8 | inference | CPU, GPU | py3 + +## Neo PyTorch (DLC) + +The following SageMaker AI Python SDK example shows how to retrieve a specific registry path. + + + from sagemaker import image_uris + image_uris.retrieve(framework='neo-pytorch',region='us-gov-east-1',version='1.6',image_scope='inference',instance_type='ml.c5.4xlarge') + +Registry path | Version | Job types (image scope) | Processor types | Python versions +---|---|---|---|--- +227234621604.dkr.ecr.us-gov-east-1.amazonaws.com/sagemaker-inference-pytorch:`<tag>` | 2.0 | inference | CPU, GPU | py3 +227234621604.dkr.ecr.us-gov-east-1.amazonaws.com/sagemaker-inference-pytorch:`<tag>` | 1.13 | inference | CPU, GPU | py3 +227234621604.dkr.ecr.us-gov-east-1.amazonaws.com/sagemaker-inference-pytorch:`<tag>` | 1.12 | inference | CPU, GPU | py3 +227234621604.dkr.ecr.us-gov-east-1.amazonaws.com/sagemaker-inference-pytorch:`<tag>` | 1.8 | inference | CPU, GPU | py3 +227234621604.dkr.ecr.us-gov-east-1.amazonaws.com/sagemaker-inference-pytorch:`<tag>` | 1.7 | inference | CPU, GPU | py3 +227234621604.dkr.ecr.us-gov-east-1.amazonaws.com/sagemaker-inference-pytorch:`<tag>` | 1.6 | inference | CPU, GPU | py3 +227234621604.dkr.ecr.us-gov-east-1.amazonaws.com/sagemaker-inference-pytorch:`<tag>` | 1.5 | inference | CPU, GPU | py3 +227234621604.dkr.ecr.us-gov-east-1.amazonaws.com/sagemaker-inference-pytorch:`<tag>` | 1.4 | inference | CPU, GPU | py3 + +## Neo Tensorflow (DLC) + +The following SageMaker AI Python SDK example shows how to retrieve a specific registry path. + + + from sagemaker import image_uris + image_uris.retrieve(framework='neo-tensorflow',region='us-gov-east-1',version='1.15.3',py_version='py3',instance_type='ml.c5.4xlarge') + +Registry path | Version | Job types (image scope) | Processor types | Python versions +---|---|---|---|---