Generation

What is generation?

Generation is a core subprocess in our SDK that focuses on producing the final output of an RAG (Retrieval-Augmented Generation) pipeline. This output typically consists of two parts:

1. Response – the synthesized answer, typically produced by the language model.

2. References – the relevant sources used to generate the response.

To support this workflow, the GL SDK provides the following features:

1. Response Synthesizer - For synthesizing the response based on the provided inputs and contexts.

2. Reference Formatter - For filtering and formatting the references in a clear, standardized format.