Fix Sagemaker Batch Endpoints (#3249)

Add different typing for @evandiewald 's heplful PR

---------

Co-authored-by: Evan Diewald <evandiewald@gmail.com>

docs/modules/models/text_embedding/examples/sagemaker-endpoint.ipynb

@@ -9,7 +9,15 @@
     "\n",
     "Let's load the SageMaker Endpoints Embeddings class. The class can be used if you host, e.g. your own Hugging Face model on SageMaker.\n",
     "\n",
-    "For instrucstions on how to do this, please see [here](https://www.philschmid.de/custom-inference-huggingface-sagemaker)"
+    "For instructions on how to do this, please see [here](https://www.philschmid.de/custom-inference-huggingface-sagemaker). **Note**: In order to handle batched requests, you will need to adjust the return line in the `predict_fn()` function within the custom `inference.py` script:\n",
+    "\n",
+    "Change from\n",
+    "\n",
+    "`return {\"vectors\": sentence_embeddings[0].tolist()}`\n",
+    "\n",
+    "to:\n",
+    "\n",
+    "`return {\"vectors\": sentence_embeddings.tolist()}`."
    ]
   },
   {
@@ -29,7 +37,7 @@
    "metadata": {},
    "outputs": [],
    "source": [
-    "from typing import Dict\n",
+    "from typing import Dict, List\n",
     "from langchain.embeddings import SagemakerEndpointEmbeddings\n",
     "from langchain.llms.sagemaker_endpoint import ContentHandlerBase\n",
     "import json\n",
@@ -39,13 +47,13 @@
     "    content_type = \"application/json\"\n",
     "    accepts = \"application/json\"\n",
     "\n",
-    "    def transform_input(self, prompt: str, model_kwargs: Dict) -> bytes:\n",
+    "    def transform_input(self, inputs: list[str], model_kwargs: Dict) -> bytes:\n",
-    "        input_str = json.dumps({\"inputs\": prompt, **model_kwargs})\n",
+    "        input_str = json.dumps({\"inputs\": inputs, **model_kwargs})\n",
     "        return input_str.encode('utf-8')\n",
     "\n",
-    "    def transform_output(self, output: bytes) -> str:\n",
+    "    def transform_output(self, output: bytes) -> List[List[float]]:\n",
     "        response_json = json.loads(output.read().decode(\"utf-8\"))\n",
-    "        return response_json[\"embeddings\"]\n",
+    "        return response_json[\"vectors\"]\n",
     "\n",
     "content_handler = ContentHandler()\n",
     "\n",

langchain/embeddings/sagemaker_endpoint.py

@@ -7,6 +7,10 @@ from langchain.embeddings.base import Embeddings
 from langchain.llms.sagemaker_endpoint import ContentHandlerBase
 
 
+class EmbeddingsContentHandler(ContentHandlerBase[List[str], List[List[float]]]):
+    """Content handler for LLM class."""
+
+
 class SagemakerEndpointEmbeddings(BaseModel, Embeddings):
     """Wrapper around custom Sagemaker Inference Endpoints.
 
@@ -62,7 +66,7 @@ class SagemakerEndpointEmbeddings(BaseModel, Embeddings):
     See: https://boto3.amazonaws.com/v1/documentation/api/latest/guide/credentials.html
     """
 
-    content_handler: ContentHandlerBase
+    content_handler: EmbeddingsContentHandler
     """The content handler class that provides an input and
     output transform functions to handle formats between LLM
     and the endpoint.
@@ -72,20 +76,20 @@ class SagemakerEndpointEmbeddings(BaseModel, Embeddings):
      Example:
         .. code-block:: python
 
-        from langchain.llms.sagemaker_endpoint import ContentHandlerBase
+        from langchain.embeddings.sagemaker_endpoint import EmbeddingsContentHandler
 
-        class ContentHandler(ContentHandlerBase):
+        class ContentHandler(EmbeddingsContentHandler):
                 content_type = "application/json"
                 accepts = "application/json"
 
-                def transform_input(self, prompt: str, model_kwargs: Dict) -> bytes:
+                def transform_input(self, prompts: List[str], model_kwargs: Dict) -> bytes:
-                    input_str = json.dumps({prompt: prompt, **model_kwargs})
+                    input_str = json.dumps({prompts: prompts, **model_kwargs})
                     return input_str.encode('utf-8')
 
-                def transform_output(self, output: bytes) -> str:
+                def transform_output(self, output: bytes) -> List[List[float]]:
                     response_json = json.loads(output.read().decode("utf-8"))
-                    return response_json[0]["generated_text"]
+                    return response_json["vectors"]
-    """
+    """  # noqa: E501
 
     model_kwargs: Optional[Dict] = None
     """Key word arguments to pass to the model."""
@@ -135,7 +139,7 @@ class SagemakerEndpointEmbeddings(BaseModel, Embeddings):
             )
         return values
 
-    def _embedding_func(self, texts: List[str]) -> List[float]:
+    def _embedding_func(self, texts: List[str]) -> List[List[float]]:
         """Call out to SageMaker Inference embedding endpoint."""
         # replace newlines, which can negatively affect performance.
         texts = list(map(lambda x: x.replace("\n", " "), texts))
@@ -179,7 +183,7 @@ class SagemakerEndpointEmbeddings(BaseModel, Embeddings):
         _chunk_size = len(texts) if chunk_size > len(texts) else chunk_size
         for i in range(0, len(texts), _chunk_size):
             response = self._embedding_func(texts[i : i + _chunk_size])
-            results.append(response)
+            results.extend(response)
         return results
 
     def embed_query(self, text: str) -> List[float]:
@@ -191,4 +195,4 @@ class SagemakerEndpointEmbeddings(BaseModel, Embeddings):
         Returns:
             Embeddings for the text.
         """
-        return self._embedding_func([text])
+        return self._embedding_func([text])[0]

langchain/llms/sagemaker_endpoint.py

@@ -1,14 +1,17 @@
 """Wrapper around Sagemaker InvokeEndpoint API."""
-from abc import ABC, abstractmethod
+from abc import abstractmethod
-from typing import Any, Dict, List, Mapping, Optional, Union
+from typing import Any, Dict, Generic, List, Mapping, Optional, TypeVar, Union
 
 from pydantic import Extra, root_validator
 
 from langchain.llms.base import LLM
 from langchain.llms.utils import enforce_stop_tokens
 
+INPUT_TYPE = TypeVar("INPUT_TYPE", bound=Union[str, List[str]])
+OUTPUT_TYPE = TypeVar("OUTPUT_TYPE", bound=Union[str, List[List[float]]])
 
-class ContentHandlerBase(ABC):
+
+class ContentHandlerBase(Generic[INPUT_TYPE, OUTPUT_TYPE]):
     """A handler class to transform input from LLM to a
     format that SageMaker endpoint expects. Similarily,
     the class also handles transforming output from the
@@ -39,9 +42,7 @@ class ContentHandlerBase(ABC):
     """The MIME type of the response data returned from endpoint"""
 
     @abstractmethod
-    def transform_input(
+    def transform_input(self, prompt: INPUT_TYPE, model_kwargs: Dict) -> bytes:
-        self, prompt: Union[str, List[str]], model_kwargs: Dict
-    ) -> bytes:
         """Transforms the input to a format that model can accept
         as the request Body. Should return bytes or seekable file
         like object in the format specified in the content_type
@@ -49,12 +50,16 @@ class ContentHandlerBase(ABC):
         """
 
     @abstractmethod
-    def transform_output(self, output: bytes) -> Any:
+    def transform_output(self, output: bytes) -> OUTPUT_TYPE:
         """Transforms the output from the model to string that
         the LLM class expects.
         """
 
 
+class LLMContentHandler(ContentHandlerBase[str, str]):
+    """Content handler for LLM class."""
+
+
 class SagemakerEndpoint(LLM):
     """Wrapper around custom Sagemaker Inference Endpoints.
 
@@ -110,7 +115,7 @@ class SagemakerEndpoint(LLM):
     See: https://boto3.amazonaws.com/v1/documentation/api/latest/guide/credentials.html
     """
 
-    content_handler: ContentHandlerBase
+    content_handler: LLMContentHandler
     """The content handler class that provides an input and
     output transform functions to handle formats between LLM
     and the endpoint.
@@ -120,7 +125,9 @@ class SagemakerEndpoint(LLM):
      Example:
         .. code-block:: python
 
-        class ContentHandler(ContentHandlerBase):
+        from langchain.llms.sagemaker_endpoint import LLMContentHandler
+
+        class ContentHandler(LLMContentHandler):
                 content_type = "application/json"
                 accepts = "application/json"