community[minor]: Adds BedrockCohereEmbeddings class + tests

[vishalseshagiri]

• Adds BedrockCohereEmbeddings class and associated tests, adds async keyword to BedrockEmbeddings.embedQuery function - Code credits to @BrianErikson (Most of the application code comes from their thread here). Some notes about the chosen design.
  • Consciously chose not to add conditions (to support Cohere embeddings) within the BedrockEmbeddings class to avoid making it brittle and unreadable.
  • Avoided extending the BedrockEmbeddings class in the BedrockCohereEmbeddings class since it'll have access to the protected _embedText function, which has no meaning in the context of the BedrockCohereEmbeddings class
• Adds missing async keyword in the BedrockEmbeddings.embedQuery function

Fixes #3315

[vercel[bot]]

The latest updates on your projects. Learn more about Vercel for Git ↗︎

Name	Status	Preview	Comments	Updated (UTC)
langchainjs-api-refs	✅ Ready (Inspect)	Visit Preview	💬 Add feedback	Apr 22, 2024 7:16am
langchainjs-docs	❌ Failed (Inspect)			Apr 22, 2024 7:16am

[jacoblee93]

@efriis this is the direction we're heading in Python right?

[vishal-oogway]

@efriis this is the direction we're heading in Python right?

Ooh looks like that's not the direction in Python https://github.com/langchain-ai/langchain/blob/master/libs/community/langchain_community/embeddings/bedrock.py#L144-L148. I can refactor it to mimic it, if that's the expectation @jacoblee93

[tomi-bigpi]

FYI: This will easily run into the 96 max document limit in the AWS API. We had to add batching logic to the calls to avoid errors.