[SageMaker] Add warning about SageMaker SDK v3 breaking changes (#2095)

* Add warning about SageMaker SDK v3 breaking changes

* Update SageMaker SDK version in notebooks to avoid breaking changes in v3

* Add missing warnings about SageMaker SDK v3 breaking changes

* Update warnings regarding SageMaker Python SDK

Author: ehcalabres

docs/hub/security-sso.md

@@ -58,21 +58,28 @@ The default value is 7 days.
 
 When enabled, Role Mapping allows you to dynamically assign [roles](./organizations-security#access-control-in-organizations) to organization members based on data provided by your Identity Provider.
 
-This section allows you to define a mapping from your IdP's user profile data from your IdP to the assigned role in Hugging Face.
+This section allows you to define a mapping from your IdP's user profile data to the assigned role in Hugging Face.
 
-- IdP Role Attribute Mapping
+- **IdP Role Attribute Path**
 
   A JSON path to an attribute in your user's IdP profile data.
+  It supports dot notation (e.g. `user.role` or `groups`).
+  For SAML, this can be a URI if your claims are URIs (e.g. `http://schemas.microsoft.com/ws/2008/06/identity/claims/role`).
 
-- Role Mapping
+- **Role Mapping**
 
   A mapping from the IdP attribute value to the assigned role in the Hugging Face organization.
+  
+  Available roles are `admin`, `write`, `contributor`, and `read`. See [roles documentation](./organizations-security#access-control-in-organizations) for more details.
 
-You must map at least one admin role.
+> [!IMPORTANT]
+> You must map at least one `admin` role in your configuration.
+
+If the attribute in the IdP response contains multiple values (e.g. a list of groups), the **first matching mapping** will be used to determine the user's role.
 
 If there is no match, a user will be assigned the default role for your organization. The default role can be customized in the `Members` section of the organization's settings.
 
-Role synchronization is performed on login.
+Role synchronization is performed on every login.
 
 #### Resource Group Mapping
 
@@ -83,13 +90,15 @@ When enabled, Resource Group Mapping allows you to dynamically assign members to
 	<img class="hidden dark:block" src="https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/enterprise/resource-group-mapping-dark.png"/>
 </div>
 
-- IdP Attribute Path
+- **IdP Attribute Path**
 
-  A JSON path to an attribute in your user's IdP profile data.
+  A JSON path to an attribute in your user's IdP profile data. Similar to Role Mapping, this supports dot notation or URIs for SAML.
+
+- **Resource Group Mapping**
 
-- Resource Group Mapping
+  A mapping from the IdP attribute value to a resource group in your Hugging Face organization. You can assign a specific role (`admin`, `write`, `contributor`, `read`) for each resource group mapping.
 
-  A mapping from the IdP attribute value to a resource group in your Hugging Face organization.
+Unlike Role Mapping, **Resource Group Mapping is additive**. If a user matches multiple mappings (e.g. they belong to multiple groups in your IdP that are mapped to different Resource Groups), they will be added to **all** matched Resource Groups.
 
 If there is no match, the user will not be assigned to any resource group.
 

docs/sagemaker/notebooks/sagemaker-sdk/deploy-embedding-models/sagemaker-notebook.ipynb

@@ -53,7 +53,15 @@
    "metadata": {},
    "outputs": [],
    "source": [
-    "!pip install \"sagemaker>=2.221.1\" --upgrade --quiet\n"
+    "!pip install \"sagemaker<3.0.0\" --upgrade --quiet"
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "metadata": {},
+   "source": [
+    "> [!WARNING]\n",
+    "> [SageMaker Python SDK v3 has been recently released](https://github.com/aws/sagemaker-python-sdk), so unless specified otherwise, all the documentation and tutorials are still using the [SageMaker Python SDK v2](https://github.com/aws/sagemaker-python-sdk/tree/master-v2). We are actively working on updating all the tutorials and examples, but in the meantime make sure to install the SageMaker SDK as `pip install \"sagemaker<3.0.0\"`."
    ]
   },
   {

docs/sagemaker/notebooks/sagemaker-sdk/deploy-llama-3-3-70b-inferentia2/sagemaker-notebook.ipynb

@@ -46,7 +46,15 @@
    "metadata": {},
    "outputs": [],
    "source": [
-    "!pip install sagemaker --upgrade --quiet"
+    "!pip install 'sagemaker<3.0.0' --upgrade --quiet"
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "metadata": {},
+   "source": [
+    "> [!WARNING]\n",
+    "> [SageMaker Python SDK v3 has been recently released](https://github.com/aws/sagemaker-python-sdk), so unless specified otherwise, all the documentation and tutorials are still using the [SageMaker Python SDK v2](https://github.com/aws/sagemaker-python-sdk/tree/master-v2). We are actively working on updating all the tutorials and examples, but in the meantime make sure to install the SageMaker SDK as `pip install \"sagemaker<3.0.0\"`."
    ]
   },
   {

docs/sagemaker/notebooks/sagemaker-sdk/evaluate-llm-lighteval/sagemaker-notebook.ipynb

@@ -25,7 +25,15 @@
    "metadata": {},
    "outputs": [],
    "source": [
-    "!pip install sagemaker --upgrade --quiet"
+    "!pip install 'sagemaker<3.0.0' --upgrade --quiet"
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "metadata": {},
+   "source": [
+    "> [!WARNING]\n",
+    "> [SageMaker Python SDK v3 has been recently released](https://github.com/aws/sagemaker-python-sdk), so unless specified otherwise, all the documentation and tutorials are still using the [SageMaker Python SDK v2](https://github.com/aws/sagemaker-python-sdk/tree/master-v2). We are actively working on updating all the tutorials and examples, but in the meantime make sure to install the SageMaker SDK as `pip install \"sagemaker<3.0.0\"`."
    ]
   },
   {

docs/sagemaker/notebooks/sagemaker-sdk/fine-tune-embedding-models/sagemaker-notebook.ipynb

@@ -39,7 +39,15 @@
    "metadata": {},
    "outputs": [],
    "source": [
-    "!pip install transformers \"datasets[s3]==2.18.0\" \"sagemaker>=2.190.0\" \"huggingface_hub[cli]\" --upgrade --quiet"
+    "!pip install transformers \"datasets[s3]==2.18.0\" \"sagemaker<3.0.0\" \"huggingface_hub[cli]\" --upgrade --quiet"
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "metadata": {},
+   "source": [
+    "> [!WARNING]\n",
+    "> [SageMaker Python SDK v3 has been recently released](https://github.com/aws/sagemaker-python-sdk), so unless specified otherwise, all the documentation and tutorials are still using the [SageMaker Python SDK v2](https://github.com/aws/sagemaker-python-sdk/tree/master-v2). We are actively working on updating all the tutorials and examples, but in the meantime make sure to install the SageMaker SDK as `pip install \"sagemaker<3.0.0\"`."
    ]
   },
   {

docs/sagemaker/source/dlcs/available.md

@@ -72,6 +72,9 @@ Let's say you want to use the training DLC for GPUs in
 
 The Python SagemMaker SDK util functions are not always up to date but it is much simpler than reconstructing the image URI yourself. 
 
+> [!WARNING]
+> [SageMaker Python SDK v3 has been recently released](https://github.com/aws/sagemaker-python-sdk), so unless specified otherwise, all the documentation and tutorials are still using the [SageMaker Python SDK v2](https://github.com/aws/sagemaker-python-sdk/tree/master-v2). We are actively working on updating all the tutorials and examples, but in the meantime make sure to install the SageMaker SDK as `pip install "sagemaker<3.0.0"`.
+
 ```python
 from sagemaker.huggingface import HuggingFaceModel, get_huggingface_llm_image_uri
 

docs/sagemaker/source/index.md

@@ -12,6 +12,9 @@ We develop new tools to simplify the adoption of custom AI accelerators like AWS
 
 By combining Hugging Face's open-source models and libraries with AWS's scalable and secure cloud services, developers can more easily and affordably incorporate advanced AI capabilities into their applications.
 
+> [!WARNING]
+> [SageMaker Python SDK v3 has been recently released](https://github.com/aws/sagemaker-python-sdk), so unless specified otherwise, all the documentation and tutorials are still using the [SageMaker Python SDK v2](https://github.com/aws/sagemaker-python-sdk/tree/master-v2). We are actively working on updating all the tutorials and examples, but in the meantime make sure to install the SageMaker SDK as `pip install "sagemaker<3.0.0"`.
+
 ## Deploy models on AWS
 
 Deploying Hugging Face models on AWS is streamlined through various services, each suited for different deployment scenarios. Here's how you can deploy your models using AWS and Hugging Face offerings.

docs/sagemaker/source/tutorials/jumpstart/jumpstart-quickstart.md

@@ -15,6 +15,9 @@ In this quickstart guide, we will deploy [Qwen/Qwen2.5-14B-Instruct](https://hug
 | SageMaker Studio domain and user profile | We recommend using SageMaker Studio for straightforward deployment and inference. Follow this [guide](https://docs.aws.amazon.com/sagemaker/latest/dg/onboard-quick-start.html). |
 | Service quotas | Most LLMs need GPU instances (e.g. ml.g5). Verify you have quota for `ml.g5.24xlarge` or [request it](https://docs.aws.amazon.com/sagemaker/latest/dg/canvas-requesting-quota-increases.html). | 
 
+> [!WARNING]
+> [SageMaker Python SDK v3 has been recently released](https://github.com/aws/sagemaker-python-sdk), so unless specified otherwise, all the documentation and tutorials are still using the [SageMaker Python SDK v2](https://github.com/aws/sagemaker-python-sdk/tree/master-v2). We are actively working on updating all the tutorials and examples, but in the meantime make sure to install the SageMaker SDK as `pip install "sagemaker<3.0.0"`.
+
 ## 2· Endpoint deployment
 
 Let's explain how you would deploy a Hugging Face model to SageMaker browsing through the Jumpstart catalog:

docs/sagemaker/source/tutorials/sagemaker-sdk/deploy-sagemaker-sdk.md

@@ -32,9 +32,12 @@ To start training locally, you need to setup an appropriate [IAM role](https://d
 Upgrade to the latest `sagemaker` version.
 
 ```bash
-pip install sagemaker --upgrade
+pip install 'sagemaker<3.0.0'
 ```
 
+> [!WARNING]
+> [SageMaker Python SDK v3 has been recently released](https://github.com/aws/sagemaker-python-sdk), so unless specified otherwise, all the documentation and tutorials are still using the [SageMaker Python SDK v2](https://github.com/aws/sagemaker-python-sdk/tree/master-v2). We are actively working on updating all the tutorials and examples, but in the meantime make sure to install the SageMaker SDK as `pip install "sagemaker<3.0.0"`.
+
 **SageMaker environment**
 
 Setup your SageMaker environment as shown below:

docs/sagemaker/source/tutorials/sagemaker-sdk/sagemaker-sdk-quickstart.md

@@ -11,9 +11,12 @@ The get started guide will show you how to quickly use Hugging Face on Amazon Sa
 Get started by installing the necessary Hugging Face libraries and SageMaker. You will also need to install [PyTorch](https://pytorch.org/get-started/locally/) if you don't already have it installed. If you run this example in SageMaker Studio, it is already installed in the notebook kernel!
 
 ```python
-pip install "sagemaker>=2.140.0" "transformers==4.26.1" "datasets[s3]==2.10.1" --upgrade
+pip install "sagemaker<3.0.0" "transformers==4.26.1" "datasets[s3]==2.10.1" --upgrade
 ```
 
+> [!WARNING]
+> [SageMaker Python SDK v3 has been recently released](https://github.com/aws/sagemaker-python-sdk), so unless specified otherwise, all the documentation and tutorials are still using the [SageMaker Python SDK v2](https://github.com/aws/sagemaker-python-sdk/tree/master-v2). We are actively working on updating all the tutorials and examples, but in the meantime make sure to install the SageMaker SDK as `pip install "sagemaker<3.0.0"`.
+
 If you want to run this example in [SageMaker Studio](https://docs.aws.amazon.com/sagemaker/latest/dg/studio.html), upgrade [ipywidgets](https://ipywidgets.readthedocs.io/en/latest/) for the 🤗 Datasets library and restart the kernel:
 
 ```python