Retrieve Passages for RAG
A 2-step recipe with Amazon Kendra. We can help you get access.
Recipe · 2 steps
Retrieve Passages for RAG
with Amazon Kendra
Retrieve passages and run a ranked query for the same question.
- Retrieve the most relevant text passages for the input question for use as grounding context in a RAG prompt.
- Run a standard ranked query for the same question to capture extractive answers and ranked document results alongside the retrieved passages.
What you'll need to have ready
- indexId: The identifier of the index to retrieve from.
- queryText: The question to retrieve passages and ranked results for.
- attributeFilter: Optional attribute filter to scope results.
- pageSize: Optional number of results per page.
For your developer
Arazzo workflow amazon-kendra:retrieve-passages-for-rag, 2 operations: Retrieve → Query. Specs on APIs.io.