Python: Add bedrock model provider parameter (#12853)

### Motivation and Context

<!-- Thank you for your contribution to the semantic-kernel repo!
Please help reviewers and future users, providing the following
information:
  1. Why is this change required?
  2. What problem does it solve?
  3. What scenario does it contribute to?
  4. If it fixes an open issue, please link to the issue here.
-->
Fixes #12833 

### Description

<!-- Describe your changes, the overall approach, the underlying design.
These notes will help understanding how your code works. Thanks! -->
Add an optional setting to the Bedrock AI connectors to allow users to
specify the Bedrock model providers (i.e. amazon, anthropic, etc) so
that they can use application inference profiles.


### Contribution Checklist

<!-- Before submitting this PR, please make sure: -->

- [x] The code builds clean without any errors or warnings
- [x] The PR follows the [SK Contribution
Guidelines](https://github.com/microsoft/semantic-kernel/blob/main/CONTRIBUTING.md)
and the [pre-submission formatting
script](https://github.com/microsoft/semantic-kernel/blob/main/CONTRIBUTING.md#development-scripts)
raises no violations
- [x] All unit tests pass, and I have added new tests where possible
- [x] I didn't break anyone 😄

---------

Co-authored-by: Dmytro Struk <[email protected]>

Author: TaoChenOSU

python/samples/concepts/setup/chat_completion_services.py

@@ -213,7 +213,7 @@ def get_bedrock_chat_completion_service_and_request_settings() -> tuple[
     """
     from semantic_kernel.connectors.ai.bedrock import BedrockChatCompletion, BedrockChatPromptExecutionSettings
 
-    chat_service = BedrockChatCompletion(service_id=service_id, model_id="anthropic.claude-3-sonnet-20240229-v1:0")
+    chat_service = BedrockChatCompletion(service_id=service_id)
     request_settings = BedrockChatPromptExecutionSettings(
         # For model specific settings, specify them in the extension_data dictionary.
         # For example, for Cohere Command specific settings, refer to:

python/semantic_kernel/connectors/ai/bedrock/README.md

@@ -38,6 +38,14 @@ bedrock_chat_completion_service = BedrockChatCompletion(runtime_client=runtime_c
 
 To find model supports by AWS regions, refer to this [AWS documentation](https://docs.aws.amazon.com/bedrock/latest/userguide/models-regions.html).
 
+### Inference profiles
+
+You can create inference profiles in AWS Bedrock to monitor and optimize the performance of your foundation models. Refer to the [AWS documentation](https://docs.aws.amazon.com/bedrock/latest/userguide/inference-profiles.html) for more information.
+
+When you are using an Application Inference Profile, you must specify the `BEDROCK_MODEL_PROVIDER` environment variable to the model provider you are using. For example, if you are using Amazon Titan, you must set `BEDROCK_MODEL_PROVIDER=amazon`. This is because an Application Inference Profile doesn't contain the model provider information, and the Bedrock connector needs to know which model provider to use so that it can create the correct request body to the Bedrock API.
+
+> An Application Inference Profile ARN is usually formatted as followed: `arn:aws:bedrock:<region>:<account-id>:application-inference-profile/<profile-id>`.
+
 ### Input & Output Modalities
 
 Foundational models in Bedrock support the multiple modalities, including text, image, and embedding. However, not all models support the same modalities. Refer to the [AWS documentation](https://docs.aws.amazon.com/bedrock/latest/userguide/models-supported.html) for more information.

python/semantic_kernel/connectors/ai/bedrock/bedrock_settings.py

@@ -2,6 +2,7 @@
 
 from typing import ClassVar
 
+from semantic_kernel.connectors.ai.bedrock.services.model_provider.bedrock_model_provider import BedrockModelProvider
 from semantic_kernel.kernel_pydantic import KernelBaseSettings
 from semantic_kernel.utils.feature_stage_decorator import experimental
 
@@ -25,10 +26,16 @@ class BedrockSettings(KernelBaseSettings):
             (Env var BEDROCK_TEXT_MODEL_ID)
         - embedding_model_id: str | None - The Amazon Bedrock embedding model ID to use.
             (Env var BEDROCK_EMBEDDING_MODEL_ID)
+        - model_provider: BedrockModelProvider | None - The Bedrock model provider to use.
+            If not provided, the model provider will be extracted from the model ID.
+            When using an Application Inference Profile where the model provider is not part
+            of the model ID, this setting must be provided.
+            (Env var BEDROCK_MODEL_PROVIDER)
     """
 
     env_prefix: ClassVar[str] = "BEDROCK_"
 
     chat_model_id: str | None = None
     text_model_id: str | None = None
     embedding_model_id: str | None = None
+    model_provider: BedrockModelProvider | None = None

python/semantic_kernel/connectors/ai/bedrock/services/bedrock_base.py

@@ -5,6 +5,7 @@
 
 import boto3
 
+from semantic_kernel.connectors.ai.bedrock.services.model_provider.bedrock_model_provider import BedrockModelProvider
 from semantic_kernel.kernel_pydantic import KernelBaseModel
 
 
@@ -19,22 +20,30 @@ class BedrockBase(KernelBaseModel, ABC):
     # Client: Use for model management
     bedrock_client: Any
 
+    bedrock_model_provider: BedrockModelProvider | None = None
+
     def __init__(
         self,
         *,
         runtime_client: Any | None = None,
         client: Any | None = None,
+        bedrock_model_provider: BedrockModelProvider | None = None,
         **kwargs: Any,
     ) -> None:
         """Initialize the Amazon Bedrock Base Class.
 
         Args:
             runtime_client: The Amazon Bedrock runtime client to use.
             client: The Amazon Bedrock client to use.
+            bedrock_model_provider: The Bedrock model provider to use.
+                If not provided, the model provider will be extracted from the model ID.
+                When using an Application Inference Profile where the model provider is not part
+                of the model ID, this setting must be provided.
             **kwargs: Additional keyword arguments.
         """
         super().__init__(
             bedrock_runtime_client=runtime_client or boto3.client("bedrock-runtime"),
             bedrock_client=client or boto3.client("bedrock"),
+            bedrock_model_provider=bedrock_model_provider,
             **kwargs,
         )

python/semantic_kernel/connectors/ai/bedrock/services/bedrock_chat_completion.py

@@ -16,6 +16,7 @@
 from semantic_kernel.connectors.ai.bedrock.bedrock_settings import BedrockSettings
 from semantic_kernel.connectors.ai.bedrock.services.bedrock_base import BedrockBase
 from semantic_kernel.connectors.ai.bedrock.services.model_provider.bedrock_model_provider import (
+    BedrockModelProvider,
     get_chat_completion_additional_model_request_fields,
 )
 from semantic_kernel.connectors.ai.bedrock.services.model_provider.utils import (
@@ -36,10 +37,7 @@
 from semantic_kernel.contents.text_content import TextContent
 from semantic_kernel.contents.utils.author_role import AuthorRole
 from semantic_kernel.contents.utils.finish_reason import FinishReason
-from semantic_kernel.exceptions.service_exceptions import (
-    ServiceInitializationError,
-    ServiceInvalidResponseError,
-)
+from semantic_kernel.exceptions.service_exceptions import ServiceInitializationError, ServiceInvalidResponseError
 from semantic_kernel.utils.async_utils import run_in_executor
 from semantic_kernel.utils.telemetry.model_diagnostics.decorators import (
     trace_chat_completion,
@@ -60,6 +58,7 @@ class BedrockChatCompletion(BedrockBase, ChatCompletionClientBase):
     def __init__(
         self,
         model_id: str | None = None,
+        model_provider: BedrockModelProvider | None = None,
         service_id: str | None = None,
         runtime_client: Any | None = None,
         client: Any | None = None,
@@ -70,6 +69,7 @@ def __init__(
 
         Args:
             model_id: The Amazon Bedrock chat model ID to use.
+            model_provider: The Bedrock model provider to use.
             service_id: The Service ID for the completion service.
             runtime_client: The Amazon Bedrock runtime client to use.
             client: The Amazon Bedrock client to use.
@@ -79,6 +79,7 @@ def __init__(
         try:
             bedrock_settings = BedrockSettings(
                 chat_model_id=model_id,
+                model_provider=model_provider,
                 env_file_path=env_file_path,
                 env_file_encoding=env_file_encoding,
             )
@@ -93,6 +94,7 @@ def __init__(
             service_id=service_id or bedrock_settings.chat_model_id,
             runtime_client=runtime_client,
             client=client,
+            bedrock_model_provider=bedrock_settings.model_provider,
         )
 
     # region Overriding base class methods
@@ -212,7 +214,7 @@ def _prepare_settings_for_request(
                 "stopSequences": settings.stop,
             }),
             "additionalModelRequestFields": get_chat_completion_additional_model_request_fields(
-                self.ai_model_id, settings
+                self.ai_model_id, settings, model_provider=self.bedrock_model_provider
             ),
         }
 

python/semantic_kernel/connectors/ai/bedrock/services/bedrock_text_completion.py

@@ -17,6 +17,7 @@
 from semantic_kernel.connectors.ai.bedrock.bedrock_settings import BedrockSettings
 from semantic_kernel.connectors.ai.bedrock.services.bedrock_base import BedrockBase
 from semantic_kernel.connectors.ai.bedrock.services.model_provider.bedrock_model_provider import (
+    BedrockModelProvider,
     get_text_completion_request_body,
     parse_streaming_text_completion_response,
     parse_text_completion_response,
@@ -41,6 +42,7 @@ class BedrockTextCompletion(BedrockBase, TextCompletionClientBase):
     def __init__(
         self,
         model_id: str | None = None,
+        model_provider: BedrockModelProvider | None = None,
         service_id: str | None = None,
         runtime_client: Any | None = None,
         client: Any | None = None,
@@ -51,6 +53,7 @@ def __init__(
 
         Args:
             model_id: The Amazon Bedrock text model ID to use.
+            model_provider: The Bedrock model provider to use.
             service_id: The Service ID for the text completion service.
             runtime_client: The Amazon Bedrock runtime client to use.
             client: The Amazon Bedrock client to use.
@@ -60,6 +63,7 @@ def __init__(
         try:
             bedrock_settings = BedrockSettings(
                 text_model_id=model_id,
+                model_provider=model_provider,
                 env_file_path=env_file_path,
                 env_file_encoding=env_file_encoding,
             )
@@ -74,6 +78,7 @@ def __init__(
             service_id=service_id or bedrock_settings.text_model_id,
             runtime_client=runtime_client,
             client=client,
+            bedrock_model_provider=bedrock_settings.model_provider,
         )
 
     # region Overriding base class methods
@@ -94,11 +99,17 @@ async def _inner_get_text_contents(
             settings = self.get_prompt_execution_settings_from_settings(settings)
         assert isinstance(settings, BedrockTextPromptExecutionSettings)  # nosec
 
-        request_body = get_text_completion_request_body(self.ai_model_id, prompt, settings)
+        request_body = get_text_completion_request_body(
+            self.ai_model_id,
+            prompt,
+            settings,
+            model_provider=self.bedrock_model_provider,
+        )
         response_body = await self._async_invoke_model(request_body)
         return parse_text_completion_response(
             self.ai_model_id,
             json.loads(response_body.get("body").read()),
+            model_provider=self.bedrock_model_provider,
         )
 
     @override
@@ -112,14 +123,20 @@ async def _inner_get_streaming_text_contents(
             settings = self.get_prompt_execution_settings_from_settings(settings)
         assert isinstance(settings, BedrockTextPromptExecutionSettings)  # nosec
 
-        request_body = get_text_completion_request_body(self.ai_model_id, prompt, settings)
+        request_body = get_text_completion_request_body(
+            self.ai_model_id,
+            prompt,
+            settings,
+            model_provider=self.bedrock_model_provider,
+        )
         response_stream = await self._async_invoke_model_stream(request_body)
         for event in response_stream.get("body"):
             chunk = event.get("chunk")
             yield [
                 parse_streaming_text_completion_response(
                     self.ai_model_id,
                     json.loads(chunk.get("bytes").decode()),
+                    model_provider=self.bedrock_model_provider,
                 )
             ]
 

python/semantic_kernel/connectors/ai/bedrock/services/bedrock_text_embedding.py

@@ -20,6 +20,7 @@
 from semantic_kernel.connectors.ai.bedrock.bedrock_settings import BedrockSettings
 from semantic_kernel.connectors.ai.bedrock.services.bedrock_base import BedrockBase
 from semantic_kernel.connectors.ai.bedrock.services.model_provider.bedrock_model_provider import (
+    BedrockModelProvider,
     get_text_embedding_request_body,
     parse_text_embedding_response,
 )
@@ -38,6 +39,7 @@ class BedrockTextEmbedding(BedrockBase, EmbeddingGeneratorBase):
     def __init__(
         self,
         model_id: str | None = None,
+        model_provider: BedrockModelProvider | None = None,
         service_id: str | None = None,
         runtime_client: Any | None = None,
         client: Any | None = None,
@@ -48,6 +50,7 @@ def __init__(
 
         Args:
             model_id: The Amazon Bedrock text embedding model ID to use.
+            model_provider: The Bedrock model provider to use.
             service_id: The Service ID for the text embedding service.
             runtime_client: The Amazon Bedrock runtime client to use.
             client: The Amazon Bedrock client to use.
@@ -57,6 +60,7 @@ def __init__(
         try:
             bedrock_settings = BedrockSettings(
                 embedding_model_id=model_id,
+                model_provider=model_provider,
                 env_file_path=env_file_path,
                 env_file_encoding=env_file_encoding,
             )
@@ -71,6 +75,7 @@ def __init__(
             service_id=service_id or bedrock_settings.embedding_model_id,
             runtime_client=runtime_client,
             client=client,
+            bedrock_model_provider=bedrock_settings.model_provider,
         )
 
     @override
@@ -87,12 +92,25 @@ async def generate_embeddings(
         assert isinstance(settings, BedrockEmbeddingPromptExecutionSettings)  # nosec
 
         results = await asyncio.gather(*[
-            self._async_invoke_model(get_text_embedding_request_body(self.ai_model_id, text, settings))
+            self._async_invoke_model(
+                get_text_embedding_request_body(
+                    self.ai_model_id,
+                    text,
+                    settings,
+                    model_provider=self.bedrock_model_provider,
+                )
+            )
             for text in texts
         ])
 
         return array([
-            array(parse_text_embedding_response(self.ai_model_id, json.loads(result.get("body").read())))
+            array(
+                parse_text_embedding_response(
+                    self.ai_model_id,
+                    json.loads(result.get("body").read()),
+                    model_provider=self.bedrock_model_provider,
+                )
+            )
             for result in results
         ])
 

python/semantic_kernel/connectors/ai/bedrock/services/model_provider/bedrock_model_provider.py

@@ -73,21 +73,34 @@ def to_model_provider(cls, model_id: str) -> "BedrockModelProvider":
 }
 
 
-def get_text_completion_request_body(model_id: str, prompt: str, settings: BedrockTextPromptExecutionSettings) -> dict:
+def get_text_completion_request_body(
+    model_id: str,
+    prompt: str,
+    settings: BedrockTextPromptExecutionSettings,
+    model_provider: BedrockModelProvider | None = None,
+) -> dict:
     """Get the request body for text completion for Amazon Bedrock models."""
-    model_provider = BedrockModelProvider.to_model_provider(model_id)
+    model_provider = model_provider or BedrockModelProvider.to_model_provider(model_id)
     return TEXT_COMPLETION_REQUEST_BODY_MAPPING[model_provider](prompt, settings)
 
 
-def parse_text_completion_response(model_id: str, response: dict) -> list[TextContent]:
+def parse_text_completion_response(
+    model_id: str,
+    response: dict,
+    model_provider: BedrockModelProvider | None = None,
+) -> list[TextContent]:
     """Parse the response from text completion for Amazon Bedrock models."""
-    model_provider = BedrockModelProvider.to_model_provider(model_id)
+    model_provider = model_provider or BedrockModelProvider.to_model_provider(model_id)
     return TEXT_COMPLETION_RESPONSE_MAPPING[model_provider](response, model_id)
 
 
-def parse_streaming_text_completion_response(model_id: str, chunk: dict) -> StreamingTextContent:
+def parse_streaming_text_completion_response(
+    model_id: str,
+    chunk: dict,
+    model_provider: BedrockModelProvider | None = None,
+) -> StreamingTextContent:
     """Parse the response from streaming text completion for Amazon Bedrock models."""
-    model_provider = BedrockModelProvider.to_model_provider(model_id)
+    model_provider = model_provider or BedrockModelProvider.to_model_provider(model_id)
     return STREAMING_TEXT_COMPLETION_RESPONSE_MAPPING[model_provider](chunk, model_id)
 
 
@@ -109,10 +122,12 @@ def parse_streaming_text_completion_response(model_id: str, chunk: dict) -> Stre
 
 
 def get_chat_completion_additional_model_request_fields(
-    model_id: str, settings: BedrockChatPromptExecutionSettings
+    model_id: str,
+    settings: BedrockChatPromptExecutionSettings,
+    model_provider: BedrockModelProvider | None = None,
 ) -> dict[str, Any] | None:
     """Get the additional model request fields for chat completion for Amazon Bedrock models."""
-    model_provider = BedrockModelProvider.to_model_provider(model_id)
+    model_provider = model_provider or BedrockModelProvider.to_model_provider(model_id)
     return CHAT_COMPLETION_ADDITIONAL_MODEL_REQUEST_FIELDS_MAPPING[model_provider](settings)
 
 
@@ -134,16 +149,23 @@ def get_chat_completion_additional_model_request_fields(
 
 
 def get_text_embedding_request_body(
-    model_id: str, text: str, settings: BedrockEmbeddingPromptExecutionSettings
+    model_id: str,
+    text: str,
+    settings: BedrockEmbeddingPromptExecutionSettings,
+    model_provider: BedrockModelProvider | None = None,
 ) -> dict:
     """Get the request body for text embedding for Amazon Bedrock models."""
-    model_provider = BedrockModelProvider.to_model_provider(model_id)
+    model_provider = model_provider or BedrockModelProvider.to_model_provider(model_id)
     return TEXT_EMBEDDING_REQUEST_BODY_MAPPING[model_provider](text, settings)
 
 
-def parse_text_embedding_response(model_id: str, response: dict) -> list[float]:
+def parse_text_embedding_response(
+    model_id: str,
+    response: dict,
+    model_provider: BedrockModelProvider | None = None,
+) -> list[float]:
     """Parse the response from text embedding for Amazon Bedrock models."""
-    model_provider = BedrockModelProvider.to_model_provider(model_id)
+    model_provider = model_provider or BedrockModelProvider.to_model_provider(model_id)
     return TEXT_EMBEDDING_RESPONSE_MAPPING[model_provider](response)