AWS sagemaker documentation change
Summary
Documentation moved to Amazon Nova 2.0 user guide, removing detailed limitations list including security control explanations about SSH access restrictions and warm pool usage
Security assessment
Changes only relocate documentation without modifying security policies. Mentions of security controls (SSH restrictions, warm pools) were removed but this appears to be part of content reorganization rather than addressing vulnerabilities
Diff
diff --git a/sagemaker/latest/dg/nova-model-limitations.md b/sagemaker/latest/dg/nova-model-limitations.md index 02901f954..53ca7ad34 100644 --- a//sagemaker/latest/dg/nova-model-limitations.md +++ b//sagemaker/latest/dg/nova-model-limitations.md @@ -7,24 +7 @@ -Amazon Nova customization doesn't support the following capabilities on SageMaker AI. - - * **SSH into the instance to find the metrics** - -Due to security controls in place, you can't SSH into the master node in the training algo-1 instance to find memory stats or NVIDIA stats and validate the training steps. - - * **Support for SageMaker trainer SDK** - -SageMaker trainer is currently not available to start training jobs. You can start training jobs only via **estimator API** today. - - * **Warm pools are not accessible to SageMaker training jobs** - -Due to security controls in place, the SageMaker warm pools can't be used to keep the instance in the warm pool till the time to live. - - * **Custom model merging** - -Merging multiple models is not currently supported. This means that creating multiple LoRA adapters and perform a multi-merge operation with the base model is not available. - - * **Supported observability tool** - -[TensorBoard](https://www.tensorflow.org/tensorboard) is the only supported observability tool to view metrics for SageMaker AI training jobs. [MLFlow](https://mlflow.org/) or [WandB](https://wandb.ai/site/) are not currently supported. For more information about using TensorBoard in SageMaker, see [TensorBoard in SageMaker AI](https://docs.aws.amazon.com//sagemaker/latest/dg/tensorboard-on-sagemaker.html). - - - +This topic has moved. For the latest information, see Limitations of customizing Amazon Nova models in the [Amazon Nova 2.0 user guide](https://docs.aws.amazon.com//nova/latest/nova2-userguide/nova-model-limitations.html). @@ -40 +17 @@ Amazon Bedrock inference -FAQs +Open weight model customization