KGI: An Integrated Framework for Knowledge Intensive Language Tasks

TL;DR

This paper introduces KGI, an integrated framework that combines state-of-the-art retrieval augmented models for knowledge-intensive tasks, enhancing accuracy through cross-examination and providing all models as open resources.

Contribution

The paper presents a unified system integrating multiple models for knowledge-intensive tasks, demonstrating improved accuracy and releasing all models used.

Findings

Enhanced dialogue accuracy via cross-examination with question answering models

Successful integration of retrieval augmented models for various knowledge tasks

Open release of all models used in the framework

Abstract

In this paper, we present a system to showcase the capabilities of the latest state-of-the-art retrieval augmented generation models trained on knowledge-intensive language tasks, such as slot filling, open domain question answering, dialogue, and fact-checking. Moreover, given a user query, we show how the output from these different models can be combined to cross-examine the outputs of each other. Particularly, we show how accuracy in dialogue can be improved using the question answering model. We are also releasing all models used in the demo as a contribution of this paper. A short video demonstrating the system is available at https://ibm.box.com/v/emnlp2022-demo.