Jumpstart and custom inference template agnostic change (#244)

* return SDK class in pytorch model.py for v1_0 and v1_1, update pytorch_create function, update unit test

* remove name and namespace from create for inference SDK to match with training SDK, functionality remains the same

* fix unit test, add metadata class usage to example notebook, remove skip test

* fix unit test again

* update integ tests

* update create call

Author: mollyheamazon

doc/cli/inference/cli_inference.md

@@ -45,10 +45,14 @@ hyp create hyp-jumpstart-endpoint [OPTIONS]
 |-----------|------|----------|-------------|
 | `--model-id` | TEXT | Yes | JumpStart model identifier (1-63 characters, alphanumeric with hyphens) |
 | `--instance-type` | TEXT | Yes | EC2 instance type for inference (must start with "ml.") |
+| `--namespace` | TEXT | No | Kubernetes namespace |
+| `--metadata-name` | TEXT | No | Name of the jumpstart endpoint object |
 | `--accept-eula` | BOOLEAN | No | Whether model terms of use have been accepted (default: false) |
 | `--model-version` | TEXT | No | Semantic version of the model (e.g., "1.0.0", 5-14 characters) |
 | `--endpoint-name` | TEXT | No | Name of SageMaker endpoint (1-63 characters, alphanumeric with hyphens) |
 | `--tls-certificate-output-s3-uri` | TEXT | No | S3 URI to write the TLS certificate (optional) |
+| `--debug` | FLAG | No | Enable debug mode (default: false) |
+
 
 ### hyp create hyp-custom-endpoint
 
@@ -70,6 +74,8 @@ hyp create hyp-custom-endpoint [OPTIONS]
 | `--image-uri` | TEXT | Yes | Docker image URI for inference |
 | `--container-port` | INTEGER | Yes | Port on which model server listens (1-65535) |
 | `--model-volume-mount-name` | TEXT | Yes | Name of the model volume mount |
+| `--namespace` | TEXT | No | Kubernetes namespace |
+| `--metadata-name` | TEXT | No | Name of the custom endpoint object |
 | `--endpoint-name` | TEXT | No | Name of SageMaker endpoint (1-63 characters, alphanumeric with hyphens) |
 | `--env` | OBJECT | No | Environment variables as key-value pairs |
 | `--metrics-enabled` | BOOLEAN | No | Enable metrics collection (default: false) |
@@ -97,6 +103,8 @@ hyp create hyp-custom-endpoint [OPTIONS]
 | `--target-value` | NUMBER | No | Target value for the CloudWatch metric |
 | `--use-cached-metrics` | BOOLEAN | No | Enable caching of metric values (default: true) |
 | `--invocation-endpoint` | TEXT | No | Invocation endpoint path (default: "invocations") |
+| `--debug` | FLAG | No | Enable debug mode (default: false) |
+
 
 ## Inference Endpoint Management Commands
 

examples/inference/SDK/inference-fsx-model-e2e.ipynb

@@ -31,6 +31,7 @@
    "source": [
     "from sagemaker.hyperpod.inference.config.hp_endpoint_config import FsxStorage, ModelSourceConfig, TlsConfig, EnvironmentVariables, ModelInvocationPort, ModelVolumeMount, Resources, Worker\n",
     "from sagemaker.hyperpod.inference.hp_endpoint import HPEndpoint\n",
+    "from sagemaker.hyperpod.common.config.metadata import Metadata\n",
     "import yaml\n",
     "import time"
    ]
@@ -42,6 +43,10 @@
    "metadata": {},
    "outputs": [],
    "source": [
+    "# If you don't set metadata name, it will be default to endpoint name\n",
+    "# If you don't set namespace, it will be default to \"default\"\n",
+    "metadata=Metadata(name='<metadata_name>', namespace='<namespace>')\n",
+    "\n",
     "tls_config=TlsConfig(tls_certificate_output_s3_uri='s3://<my-tls-bucket-name>')\n",
     "\n",
     "model_source_config = ModelSourceConfig(\n",
@@ -82,6 +87,7 @@
    "outputs": [],
    "source": [
     "fsx_endpoint = HPEndpoint(\n",
+    "    metadata=metadata,\n",
     "    endpoint_name='<my-endpoint-name>',\n",
     "    instance_type='ml.g5.8xlarge',\n",
     "    model_name='deepseek15b-fsx-test-pysdk',\n",

examples/inference/SDK/inference-jumpstart-e2e.ipynb

@@ -86,6 +86,7 @@
    "source": [
     "from sagemaker.hyperpod.inference.config.hp_jumpstart_endpoint_config import Model, Server,SageMakerEndpoint, TlsConfig, EnvironmentVariables\n",
     "from sagemaker.hyperpod.inference.hp_jumpstart_endpoint import HPJumpStartEndpoint\n",
+    "from sagemaker.hyperpod.common.config.metadata import Metadata\n",
     "import yaml\n",
     "import time"
    ]
@@ -105,6 +106,10 @@
    "metadata": {},
    "outputs": [],
    "source": [
+    "# If you don't set metadata name, it will be default to endpoint name\n",
+    "# If you don't set namespace, it will be default to \"default\"\n",
+    "metadata=Metadata(name='<metadata_name>', namespace='<namespace>')\n",
+    "\n",
     "# create configs\n",
     "model=Model(\n",
     "    model_id='deepseek-llm-r1-distill-qwen-1-5b'\n",
@@ -116,6 +121,7 @@
     "\n",
     "# create spec\n",
     "js_endpoint=HPJumpStartEndpoint(\n",
+    "    metadata=metadata,\n",
     "    model=model,\n",
     "    server=server,\n",
     "    sage_maker_endpoint=endpoint_name\n",

examples/inference/SDK/inference-s3-model-e2e.ipynb

@@ -31,6 +31,7 @@
    "source": [
     "from sagemaker.hyperpod.inference.config.hp_endpoint_config import CloudWatchTrigger, Dimensions, AutoScalingSpec, Metrics, S3Storage, ModelSourceConfig, TlsConfig, EnvironmentVariables, ModelInvocationPort, ModelVolumeMount, Resources, Worker\n",
     "from sagemaker.hyperpod.inference.hp_endpoint import HPEndpoint\n",
+    "from sagemaker.hyperpod.common.config.metadata import Metadata  \n",
     "import yaml\n",
     "import time"
    ]
@@ -42,6 +43,10 @@
    "metadata": {},
    "outputs": [],
    "source": [
+    "# If you don't set metadata name, it will be default to endpoint name\n",
+    "# If you don't set namespace, it will be default to \"default\"\n",
+    "metadata=Metadata(name='<metadata_name>', namespace='<namespace>')\n",
+    "\n",
     "tls_config=TlsConfig(tls_certificate_output_s3_uri='s3://<my-tls-bucket-name>')\n",
     "\n",
     "model_source_config = ModelSourceConfig(\n",
@@ -83,6 +88,7 @@
    "outputs": [],
    "source": [
     "s3_endpoint = HPEndpoint(\n",
+    "    metadata=metadata,\n",
     "    endpoint_name='<my-endpoint-name>',\n",
     "    instance_type='ml.g5.8xlarge',\n",
     "    model_name='deepseek15b-test-model-name',  \n",

hyperpod-custom-inference-template/hyperpod_custom_inference_template/v1_0/model.py

@@ -29,14 +29,22 @@
     CloudWatchTrigger
 )
 from sagemaker.hyperpod.inference.hp_endpoint import HPEndpoint
+from sagemaker.hyperpod.common.config.metadata import Metadata
+
 
 class FlatHPEndpoint(BaseModel):
     model_config = ConfigDict(extra="forbid")
 
+    namespace: Optional[str] = Field(
+        default="default", 
+        description="Kubernetes namespace",
+        min_length=1
+    )
+
     metadata_name: Optional[str]  = Field(
         None,
         alias="metadata_name",
-        description="Name of the jumpstart endpoint object",
+        description="Name of the custom endpoint object",
         max_length=63,
         pattern=r"^[a-zA-Z0-9](-*[a-zA-Z0-9]){0,62}$",
     )
@@ -255,7 +263,18 @@ def validate_model_source_config(self):
                 raise ValueError("fsx_file_system_id is required when model_source_type is 'fsx'")
         return self
 
+    @model_validator(mode='after')
+    def validate_name(self):
+        if not self.metadata_name and not self.endpoint_name:
+            raise ValueError("Either metadata_name or endpoint_name must be provided")
+        return self
+
     def to_domain(self) -> HPEndpoint:
+        if self.endpoint_name and not self.metadata_name:
+            self.metadata_name = self.endpoint_name
+            
+        metadata = Metadata(name=self.metadata_name, namespace=self.namespace)
+
         env_vars = None
         if self.env:
             env_vars = [
@@ -337,6 +356,7 @@ def to_domain(self) -> HPEndpoint:
             resources=resources,
         )
         return HPEndpoint(
+            metadata=metadata,
             endpoint_name=self.endpoint_name,
             instance_type=self.instance_type,
             metrics=metrics,

hyperpod-custom-inference-template/hyperpod_custom_inference_template/v1_0/schema.json

@@ -1,6 +1,20 @@
 {
   "additionalProperties": false,
   "properties": {
+    "namespace": {
+      "anyOf": [
+        {
+          "minLength": 1,
+          "type": "string"
+        },
+        {
+          "type": "null"
+        }
+      ],
+      "default": "default",
+      "description": "Kubernetes namespace",
+      "title": "Namespace"
+    },
     "metadata_name": {
       "anyOf": [
         {
@@ -13,7 +27,7 @@
         }
       ],
       "default": null,
-      "description": "Name of the jumpstart endpoint object",
+      "description": "Name of the custom endpoint object",
       "title": "Metadata Name"
     },
     "endpoint_name": {

hyperpod-jumpstart-inference-template/hyperpod_jumpstart_inference_template/v1_0/model.py

@@ -21,10 +21,17 @@
     TlsConfig
 )
 from sagemaker.hyperpod.inference.hp_jumpstart_endpoint import HPJumpStartEndpoint
+from sagemaker.hyperpod.common.config.metadata import Metadata
 
 class FlatHPJumpStartEndpoint(BaseModel):
     model_config = ConfigDict(extra="forbid")
 
+    namespace: Optional[str] = Field(
+        default="default", 
+        description="Kubernetes namespace",
+        min_length=1
+    )
+    
     accept_eula: bool = Field(
         False, alias="accept_eula", description="Whether model terms of use have been accepted"
     )
@@ -76,8 +83,18 @@ class FlatHPJumpStartEndpoint(BaseModel):
         pattern=r"^s3://([^/]+)/?(.*)$",
     )
 
+    @model_validator(mode='after')
+    def validate_name(self):
+        if not self.metadata_name and not self.endpoint_name:
+            raise ValueError("Either metadata_name or endpoint_name must be provided")
+
+
     def to_domain(self) -> HPJumpStartEndpoint:
-        # Build nested domain (pydantic) objects
+        if self.endpoint_name and not self.metadata_name:
+            self.metadata_name = self.endpoint_name
+            
+        metadata = Metadata(name=self.metadata_name, namespace=self.namespace)
+
         model = Model(
             accept_eula=self.accept_eula,
             model_id=self.model_id,
@@ -91,6 +108,7 @@ def to_domain(self) -> HPJumpStartEndpoint:
             TlsConfig(tls_certificate_output_s3_uri=self.tls_certificate_output_s3_uri)
         )
         return HPJumpStartEndpoint(
+            metadata=metadata,
             model=model,
             server=server,
             sage_maker_endpoint=sage_ep,

hyperpod-jumpstart-inference-template/hyperpod_jumpstart_inference_template/v1_0/schema.json

@@ -1,6 +1,20 @@
 {
   "additionalProperties": false,
   "properties": {
+    "namespace": {
+      "anyOf": [
+        {
+          "minLength": 1,
+          "type": "string"
+        },
+        {
+          "type": "null"
+        }
+      ],
+      "default": "default",
+      "description": "Kubernetes namespace",
+      "title": "Namespace"
+    },
     "accept_eula": {
       "default": false,
       "description": "Whether model terms of use have been accepted",

src/sagemaker/hyperpod/cli/commands/inference.py

@@ -20,50 +20,38 @@
 
 # CREATE
 @click.command("hyp-jumpstart-endpoint")
-@click.option(
-    "--namespace",
-    type=click.STRING,
-    required=False,
-    default="default",
-    help="Optional. The namespace of the jumpstart model endpoint to create. Default set to 'default'",
-)
 @click.option("--version", default="1.0", help="Schema version to use")
+@click.option("--debug", default=False, help="Enable debug mode")
 @generate_click_command(
     schema_pkg="hyperpod_jumpstart_inference_template",
     registry=JS_REG,
 )
 @_hyperpod_telemetry_emitter(Feature.HYPERPOD_CLI, "create_js_endpoint_cli")
 @handle_cli_exceptions()
-def js_create(name, namespace, version, js_endpoint):
+def js_create(version, debug, js_endpoint):
     """
     Create a jumpstart model endpoint.
     """
-
-    js_endpoint.create(name=name, namespace=namespace)
+    click.echo(f"Using version: {version}")
+    js_endpoint.create(debug=debug)
 
 
 @click.command("hyp-custom-endpoint")
-@click.option(
-    "--namespace",
-    type=click.STRING,
-    required=False,
-    default="default",
-    help="Optional. The namespace of the jumpstart model endpoint to create. Default set to 'default'",
-)
 @click.option("--version", default="1.0", help="Schema version to use")
+@click.option("--debug", default=False, help="Enable debug mode")
 @generate_click_command(
     schema_pkg="hyperpod_custom_inference_template",
     registry=C_REG,
 )
 @_hyperpod_telemetry_emitter(Feature.HYPERPOD_CLI, "create_custom_endpoint_cli")
 @handle_cli_exceptions()
-def custom_create(name, namespace, version, custom_endpoint):
+def custom_create(version, debug, custom_endpoint):
     """
     Create a custom model endpoint.
     """
-
-    custom_endpoint.create(name=name, namespace=namespace)
-
+    click.echo(f"Using version: {version}")
+    custom_endpoint.create(debug=debug)
+    
 
 # INVOKE
 @click.command("hyp-custom-endpoint")

src/sagemaker/hyperpod/cli/commands/init.py

@@ -393,11 +393,7 @@ def _default_create(region):
                 filtered_config = _filter_cli_metadata_fields(data)
                 flat = model(**filtered_config)
                 domain = flat.to_domain()
-                # TODO: update inference SDK to include name and namespace in the call
-                if template == "hyp-custom-endpoint" or template == "hyp-jumpstart-endpoint":
-                    domain.create(namespace=namespace)
-                elif template == "hyp-pytorch-job":
-                    domain.create()
+                domain.create()
 
 
     except Exception as e: