SagemakerGenerator

Generate text using large language models deployed on Amazon Sagemaker.

Key Features

• Connects to LLMs hosted on Amazon SageMaker Inference Endpoints.
• Supports custom model attributes, required for models like Llama 2 (for example, accept_eula: True).
• Configurable generation parameters like max_new_tokens, temperature, and top_p.
• Uses AWS credentials from environment variables or explicit configuration.
• Returns generated text strings and response metadata.

Configuration

Authentication

Connect deepset AI Platform with Amazon Bedrock first.

For detailed explanation, see Use Amazon Bedrock and SageMaker Models.

1. Drag the SagemakerGenerator component onto the canvas from the Component Library.
2. Click the component to open the configuration panel.
3. On the General tab:
   1. Enter the SageMaker Model Endpoint name, such as jumpstart-dft-meta-textgenerationneuron-llama-2-7b.
4. Go to the Advanced tab to configure AWS credentials, custom attributes, and generation kwargs.

Connections

SagemakerGenerator receives a prompt string from PromptBuilder. It outputs replies (a list of generated strings) and meta (response metadata). Connect its replies output to AnswerBuilder or DeepsetAnswerBuilder.

Usage Example

This is a RAG pipeline that uses SagemakerGenerator with a Llama2 model:

components:
  bm25_retriever: # Selects the most similar documents from the document store
    type: haystack_integrations.components.retrievers.opensearch.bm25_retriever.OpenSearchBM25Retriever
    init_parameters:
      document_store:
        type: haystack_integrations.document_stores.opensearch.document_store.OpenSearchDocumentStore
        init_parameters:
          hosts:
          index: 'Standard-Index-English'
          max_chunk_bytes: 104857600
          embedding_dim: 768
          return_embedding: false
          method:
          mappings:
          settings:
          create_index: true
          http_auth:
          use_ssl:
          verify_certs:
          timeout:
      top_k: 20 # The number of results to return
      fuzziness: 0

  query_embedder:
    type: deepset_cloud_custom_nodes.embedders.nvidia.text_embedder.DeepsetNvidiaTextEmbedder
    init_parameters:
      normalize_embeddings: true
      model: intfloat/e5-base-v2

  embedding_retriever: # Selects the most similar documents from the document store
    type: haystack_integrations.components.retrievers.opensearch.embedding_retriever.OpenSearchEmbeddingRetriever
    init_parameters:
      document_store:
        type: haystack_integrations.document_stores.opensearch.document_store.OpenSearchDocumentStore
        init_parameters:
          hosts:
          index: 'Standard-Index-English'
          max_chunk_bytes: 104857600
          embedding_dim: 768
          return_embedding: false
          method:
          mappings:
          settings:
          create_index: true
          http_auth:
          use_ssl:
          verify_certs:
          timeout:
      top_k: 20 # The number of results to return

  document_joiner:
    type: haystack.components.joiners.document_joiner.DocumentJoiner
    init_parameters:
      join_mode: concatenate

  ranker:
    type: deepset_cloud_custom_nodes.rankers.nvidia.ranker.DeepsetNvidiaRanker
    init_parameters:
      model: intfloat/simlm-msmarco-reranker
      top_k: 8

  meta_field_grouping_ranker:
    type: haystack.components.rankers.meta_field_grouping_ranker.MetaFieldGroupingRanker
    init_parameters:
      group_by: file_id
      subgroup_by:
      sort_docs_by: split_id

  prompt_builder:
    type: haystack.components.builders.prompt_builder.PromptBuilder
    init_parameters:
      template: |-
        You are a technical expert.
        You answer questions truthfully based on provided documents.
        If the answer exists in several documents, summarize them.
        Ignore documents that don't contain the answer to the question.
        Only answer based on the documents provided. Don't make things up.
        If no information related to the question can be found in the document, say so.
        Always use references in the form [NUMBER OF DOCUMENT] when using information from a document, e.g. [3] for Document [3] .
        Never name the documents, only enter a number in square brackets as a reference.
        The reference must only refer to the number that comes in square brackets after the document.
        Otherwise, do not use brackets in your answer and reference ONLY the number of the document without mentioning the word document.

        These are the documents:
        {%- if documents|length > 0 %}
        {%- for document in documents %}
        Document [{{ loop.index }}] :
        Name of Source File: {{ document.meta.file_name }}
        {{ document.content }}
        {% endfor -%}
        {%- else %}
        No relevant documents found.
        Respond with "Sorry, no matching documents were found, please adjust the filters or try a different question."
        {% endif %}

        Question: {{ question }}
        Answer:

      required_variables: "*"
  answer_builder:
    type: deepset_cloud_custom_nodes.augmenters.deepset_answer_builder.DeepsetAnswerBuilder
    init_parameters:
      reference_pattern: acm

  LangfuseConnector:
    type: haystack_integrations.components.connectors.langfuse.langfuse_connector.LangfuseConnector
    init_parameters:
      name: RAG-QA-Claude-3.5-Sonnet-en
      public: false
      public_key:
        type: env_var
        env_vars:
        - LANGFUSE_PUBLIC_KEY
        strict: false
      secret_key:
        type: env_var
        env_vars:
        - LANGFUSE_SECRET_KEY
        strict: false
      httpx_client:
      span_handler:
  SagemakerGenerator:
    type: haystack_integrations.components.generators.amazon_sagemaker.sagemaker.SagemakerGenerator
    init_parameters:
      model: jumpstart-dft-meta-textgenerationneuron-llama-2-7b
      aws_access_key_id:
        type: env_var
        env_vars:
        - AWS_ACCESS_KEY_ID
        strict: false
      aws_secret_access_key:
        type: env_var
        env_vars:
        - AWS_SECRET_ACCESS_KEY
        strict: false
      aws_session_token:
        type: env_var
        env_vars:
        - AWS_SESSION_TOKEN
        strict: false
      aws_region_name:
        type: env_var
        env_vars:
        - AWS_DEFAULT_REGION
        strict: false
      aws_profile_name:
        type: env_var
        env_vars:
        - AWS_PROFILE
        strict: false
      aws_custom_attributes:
      - accept_eula: true
      generation_kwargs:

connections:  # Defines how the components are connected
- sender: bm25_retriever.documents
  receiver: document_joiner.documents
- sender: query_embedder.embedding
  receiver: embedding_retriever.query_embedding
- sender: embedding_retriever.documents
  receiver: document_joiner.documents
- sender: document_joiner.documents
  receiver: ranker.documents
- sender: ranker.documents
  receiver: meta_field_grouping_ranker.documents
- sender: meta_field_grouping_ranker.documents
  receiver: prompt_builder.documents
- sender: meta_field_grouping_ranker.documents
  receiver: answer_builder.documents
- sender: prompt_builder.prompt
  receiver: answer_builder.prompt
- sender: prompt_builder.prompt
  receiver: SagemakerGenerator.prompt
- sender: SagemakerGenerator.replies
  receiver: answer_builder.replies

inputs:  # Define the inputs for your pipeline
  query:  # These components will receive the query as input
  - "bm25_retriever.query"
  - "query_embedder.text"
  - "ranker.query"
  - "prompt_builder.question"
  - "answer_builder.query"

  filters:  # These components will receive a potential query filter as input
  - "bm25_retriever.filters"
  - "embedding_retriever.filters"

outputs:  # Defines the output of your pipeline
  documents: "meta_field_grouping_ranker.documents"  # The output of the pipeline is the retrieved documents
  answers: "answer_builder.answers"  # The output of the pipeline is the generated answers

max_runs_per_component: 100

metadata: {}

Parameters

Inputs

Parameter	Type	Default	Description
prompt	str		The prompt with instructions for the model.
generation_kwargs	Optional[Dict[str, Any]]	None	Additional keyword arguments for text generation. These parameters potentially override the parameters passed in pipeline configuration.

Outputs

Parameter	Type	Default	Description
replies	List[str]		A list of strings containing the generated responses.
meta	List[Dict[str, Any]]		A list of dictionaries containing the metadata for each response.

Init Parameters

These are the parameters you can configure in Pipeline Builder:

Parameter	Type	Default	Description
aws_access_key_id	Optional[Secret]	Secret.from_env_var(['AWS_ACCESS_KEY_ID'], strict=False)	The Secret for AWS access key ID.
aws_secret_access_key	Optional[Secret]	Secret.from_env_var(['AWS_SECRET_ACCESS_KEY'], strict=False)	The Secret for AWS secret access key.
aws_session_token	Optional[Secret]	Secret.from_env_var(['AWS_SESSION_TOKEN'], strict=False)	The Secret for AWS session token.
aws_region_name	Optional[Secret]	Secret.from_env_var(['AWS_DEFAULT_REGION'], strict=False)	The Secret for AWS region name. If not provided, the default region will be used.
aws_profile_name	Optional[Secret]	Secret.from_env_var(['AWS_PROFILE'], strict=False)	The Secret for AWS profile name. If not provided, the default profile will be used.
model	str		The name for SageMaker Model Endpoint.
aws_custom_attributes	Optional[Dict[str, Any]]	None	Custom attributes to be passed to SageMaker, for example {"accept_eula": True} in case of Llama-2 models.
generation_kwargs	Optional[Dict[str, Any]]	None	Additional keyword arguments for text generation. For a list of supported parameters, see your model's documentation. Llama-2 models support: max_new_tokens, temperature, top_p, return_full_text.

Run Method Parameters

These are the parameters you can configure for the component's run() method. This means you can pass these parameters at query time through the API, in Playground, or when running a job. For details, see Modify Pipeline Parameters at Query Time.

Parameter	Type	Default	Description
prompt	str		The string prompt to use for text generation.
generation_kwargs	Optional[Dict[str, Any]]	None	Additional keyword arguments for text generation. These parameters will potentially override the parameters passed in the __init__ method.