​
Node Details

  • Name: HydeRetriever
  • Type: HydeRetriever
  • Version: 3.0
  • Category: Retrievers
  • Base Classes: HydeRetriever, BaseRetriever

​
Input Parameters

  1. Language Model (required)
    • Type: BaseLanguageModel
    • Description: The language model used to generate hypothetical documents.
  2. Vector Store (required)
    • Type: VectorStore
    • Description: The vector store to retrieve documents from.
  3. Query (optional)
    • Type: string
    • Description: Specific query to retrieve documents. If not provided, the user’s question will be used.
  4. Select Defined Prompt (required)
    • Type: options
    • Description: Pre-defined prompt templates for different use cases.
    • Options: websearch, scifact, arguana, trec-covid, fiqa, dbpedia-entity, trec-news, mr-tydi
    • Default: websearch
  5. Custom Prompt (optional)
    • Type: string
    • Description: A custom prompt template that overrides the defined prompt if provided.
  6. Top K (optional)
    • Type: number
    • Description: Number of top results to fetch.
    • Default: 4

​
Outputs

  1. HyDE Retriever
    • Type: HydeRetriever
    • Description: The configured HyDE Retriever object.
  2. Document
    • Type: Document[]
    • Description: An array of retrieved document objects containing metadata and page content.
  3. Text
    • Type: string
    • Description: Concatenated string of page content from all retrieved documents.

​
Usage

The HyDE Retriever is particularly useful in scenarios where traditional keyword-based retrieval might fall short. It’s effective for:
  • Answering complex questions that require contextual understanding
  • Retrieving documents for topics with limited or ambiguous keywords
  • Improving retrieval performance in domain-specific applications
By generating a hypothetical answer or passage, the HyDE Retriever can capture the semantic intent of the query more effectively, leading to more relevant document retrieval.

​
Note

The performance of the HyDE Retriever heavily depends on the quality of the language model and the appropriateness of the prompt template for the given task. Experimenting with different prompts and fine-tuning the language model for specific domains can significantly improve retrieval results.