[Amazon Bedrock] Add AmazonBedrockEmbedder

[julian-risch]

In addition to the recently merged AmazonBedrockGenerator #153, we need an AmazonBedrockEmbedder for the embedding models.

Tasks

Preview Give feedback

1. The code is documented with docstrings and was merged in the main branch
2. The package has been released on PyPI
3. Docs are published at https://docs.haystack.deepset.ai/ (in review...)
4. integration tile update - https://github.com/deepset-ai/haystack-integrations (in review...)

[lambda-science]

Just a heads up for anyone really wanting to use AWS Bedrock Embedding with Haystack 2.0 but can't because it's not implemented yet. It's possible to be done in a hacky-way with LiteLLM Proxy https://litellm.vercel.app/docs/simple_proxy

If you have a docker-compose hosting your Haystack API you can do something like this. You can also do it with this sample configuration:
docker-compose.yml:

  litellm:
    image: ghcr.io/berriai/litellm:main-v1.17.2
    ports:
      - "8001:8001"
    environment:
      - AWS_PROFILE=${AWS_PROFILE_NAME}
      - AWS_REGION=${AWS_REGION_NAME}
      - LITELLM_LOG=INFO
    volumes:
      - ./litellm-config.yaml:/app/config.yaml
      - ~/.aws/:/root/.aws/
    command: [ "--config", "/app/config.yaml", "--port", "8001", "--num_workers", "8" ]

litellm-config.yaml

model_list:
  - model_name: bedrock-titan-embed-text-v1
    litellm_params:
      model: bedrock/amazon.titan-embed-text-v1

litellm_settings:
  drop_params: True

And in your Haystack 2.x pipeline:

from haystack.components.embedders import OpenAIDocumentEmbedder
from haystack.components.embedders import OpenAITextEmbedder
API_BASE_URL = "http://litellm:8001"
document_embedder = OpenAIDocumentEmbedder(api_base_url=API_BASE_URL, model_name="bedrock-titan-embed-text-v1")
query_embedder = OpenAITextEmbedder(api_base_url=API_BASE_URL, model_name="bedrock-titan-embed-text-v1")

Not optimal because it requeries an external container but it's a "quick win" as we say ahah
I'm not sure I have the programming skills to do a proper implementation in Haystack of Bedrock embedding unfortunately :/

[dfokina]

@anakin87 The docs are published: https://docs.haystack.deepset.ai/v2.0/docs/amazonbedrockdocumentembedder
https://docs.haystack.deepset.ai/v2.0/docs/amazonbedrocktextembedder