fabricatio_rag.capabilities.enrich

LLM-driven text enrichment: generate question-answer pairs from source text.

Classes

EnrichChunkText

Generate question-answer pairs from text chunks via LLM enrichment.

Module Contents

class fabricatio_rag.capabilities.enrich.EnrichChunkText(/, **data: Any)

Bases: fabricatio_core.capabilities.propose.Propose, abc.ABC

Generate question-answer pairs from text chunks via LLM enrichment.

async enrich(enrich_guideline: str, chunk: str) → fabricatio_rag.models.qa.EnrichmentResult
async enrich(enrich_guideline: str, chunk: List[str]) → List[fabricatio_rag.models.qa.EnrichmentResult]

Generate QAPairs from text chunk(s) guided by enrichment instructions.

Parameters:

  • enrich_guideline – Free-form NL describing what kinds of questions to generate (e.g. “factual recall”, “conceptual”, “multi-hop”).

  • chunk – Single string or list of strings to enrich.

Returns:

an EnrichmentResult with qa_pairs for that chunk. For a list of str: a list of EnrichmentResults, one per input chunk, preserving input order.

Return type:

For a single str