Skip to main content

Interface: HydeConfig

Defined in: packages/agentos/src/rag/HydeRetriever.ts:24

Properties​

adaptiveThreshold?​

optional adaptiveThreshold: boolean

Defined in: packages/agentos/src/rag/HydeRetriever.ts:34

Use adaptive thresholding (step down when no results). Default: true.

enabled?​

optional enabled: boolean

Defined in: packages/agentos/src/rag/HydeRetriever.ts:26

Enable HyDE retrieval. Default: false.

fullAnswerGranularity?​

optional fullAnswerGranularity: boolean

Defined in: packages/agentos/src/rag/HydeRetriever.ts:40

Use full-answer granularity (recommended by research). Default: true.

hypothesisCount?​

optional hypothesisCount: number

Defined in: packages/agentos/src/rag/HydeRetriever.ts:55

Number of diverse hypothetical documents to generate per query.

Multi-hypothesis HyDE generates N hypotheses from different perspectives (technical, practical/example, overview) and searches with each embedding. Results are deduplicated by chunk ID, keeping the highest score.

Higher values improve recall at the cost of additional LLM calls.

  • 1: Original single-hypothesis HyDE (fastest)
  • 3: Recommended default (good diversity/cost tradeoff)
  • 5: Maximum diversity (highest recall, most expensive)

Default: 3.

hypothesisSystemPrompt?​

optional hypothesisSystemPrompt: string

Defined in: packages/agentos/src/rag/HydeRetriever.ts:38

Custom system prompt for hypothesis generation.

initialThreshold?​

optional initialThreshold: number

Defined in: packages/agentos/src/rag/HydeRetriever.ts:28

Initial similarity threshold. Default: 0.7.

maxHypothesisTokens?​

optional maxHypothesisTokens: number

Defined in: packages/agentos/src/rag/HydeRetriever.ts:36

Max tokens for hypothesis generation. Default: 200.

minThreshold?​

optional minThreshold: number

Defined in: packages/agentos/src/rag/HydeRetriever.ts:30

Minimum threshold to step down to. Default: 0.3.

thresholdStep?​

optional thresholdStep: number

Defined in: packages/agentos/src/rag/HydeRetriever.ts:32

Step size for adaptive thresholding. Default: 0.1.