Documentation

Result Summarizer Transformer

This LLM-based summarizer is the center piece of Retrieval Augmented Generations. It uses the found search results to generate an answer to the input query.

image-20240811-161731.png

This summarizer is based on Open AI GPT and has the following configuration parameters

  • GPT Endpoint is determines the endpoint for a deployed GPT on Azure OpenAI in the format https://<baseUrl>>.openai.azure.com/openai/deployments/<deploymentName>

  • Password is the (admin) API key as provided in your Azure OpenAI configuration

  • Fieldname to summarize: defines which field should be used from the response to generate the answer

  • Number of results to combine in one response: defines how many of the top X results should be used to generate the answer