​
Available Components

AWS Bedrock Knowledge Base Retriever

Connects to AWS Bedrock Knowledge Base for retrieving information from large knowledge bases

Cohere Rerank Retriever

Utilizes Cohere’s reranking capabilities to improve document relevance

Custom Retriever

Allows customization of retrieval output format, including both document content and metadata

Embeddings Filter Retriever

Uses embeddings to filter out unrelated documents, improving retrieval relevance

Extract Metadata Retriever

Extracts metadata from documents for better retrieval

HyDE Retriever

Leverages Hypothetical Document Embeddings to improve retrieval performance

LLM Filter Retriever

Employs a language model to extract only relevant content from initially retrieved documents

Multi Query Retriever

Generates multiple queries from different perspectives to overcome limitations

Prompt Retriever

Stores and manages prompt templates for dynamic selection in conversational AI systems

Reciprocal Rank Fusion Retriever

Uses the RRF algorithm to re-rank search results from multiple query generations

Similarity Score Threshold Retriever

Filters and retrieves documents based on a minimum similarity percentage

Vector Store Retriever

A general-purpose retriever for storing and querying vector representations of data

Voyage AI Rerank Retriever

Enhances retrieval using Voyage AI’s reranking models for better semantic relevance
These retrievers are beneficial for various use cases, including:
  • Enhancing search relevance in document retrieval systems
  • Improving question-answering capabilities by providing more accurate and diverse context
  • Refining information retrieval for research and analysis tasks
  • Customizing retrieval processes for specific domain requirements
  • Integrating with large-scale knowledge bases and AI services
By leveraging these specialized retrievers, developers can significantly improve the quality and relevance of retrieved information in their AI-powered applications.