`_deploy_for_ic` passes `instance_type` twice to `_deploy()` causing `TypeError` when deploying `CustomOrchestrator` as Inference Component

从 sagemaker-serve/src/sagemaker/serve/model_builder.py 中的 _deploy_for_ic() 以及 6189-6196 行附近的 deploy() 调用开始。复现文档中所述的 CustomOrchestrator 部署，然后检查 instance_type 和 initial_instance_count 如何传递到 _deploy()。当 deploy() 能够在不出现重复关键字 TypeError 的情况下创建推理组件时，即表示完成。

PySDK Version

• PySDK V2 (2.x)
• PySDK V3 (3.x)

Describe the bug

When deploying a CustomOrchestrator as an Inference Component — as documented in the Build and deploy AI inference workflows with new enhancements to the Amazon SageMaker Python SDK blog post and the Llama3.1-Mistral reference notebook — calling deploy() raises a TypeError: got multiple values for keyword argument 'instance_type'.

The issue is in _deploy_for_ic() (model_builder.py L4227-4237): instance_type and initial_instance_count are passed both as explicit keyword arguments and via **kwargs spread to self._deploy().

To reproduce

from sagemaker.serve.model_builder import ModelBuilder, SchemaBuilder
from sagemaker.serve.spec.inference_base import CustomOrchestrator
from sagemaker.core.inference_config import ResourceRequirements
from sagemaker.core.helper.session_helper import Session, get_execution_role

class MyOrchestrator(CustomOrchestrator):
    def __init__(self, endpoint_name, component_names):
        super().__init__()
        self.endpoint_name = endpoint_name
        self.component_names = component_names

    def handle(self, data, context=None):
        import json
        response = self.client.invoke_endpoint(
            EndpointName=self.endpoint_name,
            InferenceComponentName=self.component_names[0],
            Body=data if isinstance(data, (str, bytes)) else json.dumps(data),
            ContentType="application/json"
        )
        return json.loads(response["Body"].read())

role = get_execution_role()
sess = Session()

# Step 1: Build the orchestrator
orchestrator = ModelBuilder(
    inference_spec=MyOrchestrator(
        endpoint_name="my-existing-endpoint",
        component_names=["base-ic", "adapter-ic"],
    ),
    dependencies={"auto": False, "custom": ["cloudpickle"]},
    sagemaker_session=sess,
    role_arn=role,
    schema_builder=SchemaBuilder(sample_input="Test", sample_output={"generated_text": "test"}),
)

# Workaround for missing constructor fields (separate issue)
orchestrator.resource_requirements = ResourceRequirements(
    requests={"memory": 4096, "num_accelerators": 1, "copies": 1, "num_cpus": 2}
)
orchestrator.inference_component_name = "my-orchestrator-ic"

orchestrator.build()

# Step 2: Deploy — this triggers the bug
orchestrator.deploy(
    endpoint_name="my-existing-endpoint",
    custom_orchestrator_instance_type="ml.g6.12xlarge",
    initial_instance_count=1,
)

Expected behavior

deploy() should deploy the CustomOrchestrator as an Inference Component on the specified endpoint without error.

Screenshots or logs

│   4226 │   │   │   # Create new IC via _deploy()                                                 │
│ ❱ 4227 │   │   │   return self._deploy(                                                          │
│   4228 │   │   │   │   built_model=built_model,                                                  │
│   4229 │   │   │   │   endpoint_name=endpoint_name,                                              │
│   4230 │   │   │   │   endpoint_type=EndpointType.INFERENCE_COMPONENT_BASED,                     │
╰──────────────────────────────────────────────────────────────────────────────────────────────────╯
TypeError: sagemaker.serve.model_builder.ModelBuilder._deploy() got multiple values for keyword argument 
'instance_type'

Full traceback:

/opt/conda/lib/python3.12/site-packages/sagemaker/serve/model_builder.py:6189 in deploy
│ ❱ 6189 │   │   │   │   │   │   self._deploy_for_ic(
│   6190 │   │   │   │   │   │   │   ic_data=custom_orchestrator,
│   6191 │   │   │   │   │   │   │   container_timeout_in_seconds=container_timeout_in_seconds,
│   6192 │   │   │   │   │   │   │   instance_type=custom_orchestrator_instance_type or instance_type,

/opt/conda/lib/python3.12/site-packages/sagemaker/serve/model_builder.py:4227 in _deploy_for_ic
│ ❱ 4227 │   │   │   return self._deploy(
│   4228 │   │   │   │   built_model=built_model,
│   4229 │   │   │   │   endpoint_name=endpoint_name,
│   4230 │   │   │   │   endpoint_type=EndpointType.INFERENCE_COMPONENT_BASED,

TypeError: sagemaker.serve.model_builder.ModelBuilder._deploy() got multiple values for keyword argument 'instance_type'

System information

• SageMaker Python SDK version: sagemaker-serve 1.20.0 (SDK V3)
• Framework name: SageMaker Distribution (SMD) container
• Framework version: sagemaker-distribution-prod:3.2.0-cpu
• Python version: 3.12
• CPU or GPU: GPU (ml.g6.2xlarge endpoint)
• Custom Docker image (Y/N): N

Additional context

Root cause analysis:

In deploy() (L6189-6196), _deploy_for_ic is called with instance_type as an explicit kwarg:

self._deploy_for_ic(
    ic_data=custom_orchestrator,
    container_timeout_in_seconds=container_timeout_in_seconds,
    instance_type=custom_orchestrator_instance_type or instance_type,  # explicit
    initial_instance_count=custom_orchestrator_initial_instance_count or initial_instance_count,  # explicit
    endpoint_name=endpoint_name,
    **kwargs,
)

Then in _deploy_for_ic() (L4227-4237):

def _deploy_for_ic(self, ic_data, endpoint_name, **kwargs):
    ...
    return self._deploy(
        built_model=built_model,
        endpoint_name=endpoint_name,
        endpoint_type=EndpointType.INFERENCE_COMPONENT_BASED,
        resources=resource_requirements,
        inference_component_name=ic_name,
        instance_type=kwargs.get("instance_type", self.instance_type),  # extracted from kwargs
        initial_instance_count=kwargs.get("initial_instance_count", 1),  # extracted from kwargs
        **kwargs,  # ← kwargs STILL contains instance_type → duplicate!
    )

instance_type is extracted from kwargs on one line, then **kwargs is spread on the next — passing the same key twice to _deploy().