Reason first, then respond: Modular Generation for Knowledge-infused Dialogue

By Leonard Adolphs, Kurt Shuster, Jack Urbanek, Arthur Szlam, Jason Aaron Edward Weston. 2021

43 citations recorded by the source corpus.

Large language models can produce fluent dialogue but often hallucinate factual inaccuracies. While retrieval-augmented models help alleviate this issue, they still face a difficult challenge of both reasoning to provide correct knowledge and generating conversation simultaneously. In this work, we propose a modular model, Knowledge to Response (K2R), for incorporating knowledge into conversational agents, which breaks down this problem into two easier steps. K2R first generates a knowledge sequence, given a dialogue context, as an intermediate step. After this"reasoning step", the model then attends to its own generated knowledge sequence, as well as the dialogue context, to produce a final response. In detailed experiments, we find that such a model hallucinates less in knowledge-grounded dialogue tasks, and has advantages in terms of interpretability and modularity. In particular, it can be used to fuse QA and dialogue systems together to enable dialogue agents to give knowledgeable answers, or QA models to give conversational responses in a zero-shot setting.

Subfields: Artificial Intelligence, Natural Language Processing, Prompting & Reasoning

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Artificial Intelligence

In Natural Language Processing

In Prompting & Reasoning

Citations in the corpus

Language Models that Seek for Knowledge: Modular Search & Generation for Dialogue and Prompt Completion (2022, methodology): “For NQ, we use three different settings: with all documents as input, with only the gold document, and with a sampled dialogue history context, following (Adolphs et al., 2021).”
Improving alignment of dialogue agents via targeted human judgements (2022, methodology): “SeeKeR (Adolphs et al., 2021; Shuster et al., 2022a) uses an intermediate knowledge-extraction step to reduce from several documents concatenated together to a smaller knowledge sentence, while we rely on search engine text snippets.”
BlenderBot 3: a deployed conversational agent that continually learns to responsibly engage (2022, methodology): “The new FITS dataset (Xu et al., 2022b) is available at https://www.parl.ai/projects/fits, the SafetyMix benchmark at https://www.parl.ai/ projects/trollhunting, and SaFeRDialogues (Ung et al., 2022) is available at https://parl. ai/projects/saferdialogues.”
Human Evaluation of Conversations is an Open Problem: comparing the sensitivity of various methods for evaluating dialogue agents (2022, background): “However, this no longer as easily suffices for free-form, conversational and long-form QA where answers are more open-ended (Fan et al., 2019; Adolphs et al., 2021).”
Learning New Skills after Deployment: Improving open-domain internet-driven dialogue with human feedback (2022, methodology): “• SeeKeR (Shuster et al., 2022a); again uses a similar 2.7B parameter architecture, but utilizing the Knowledge-to-response (K2R) approach (Adolphs et al., 2021) which performs a multi-step generation procedure: ﬁrst generating a relevant knowledge response, and then conditions on that to generate…”
Chain-of-Verification Reduces Hallucination in Large Language Models (2023, background): “…tasks can be gained by encouraging language models to first generate internal thoughts or reasoning chains before responding (Wei et al., 2022; Adolphs et al., 2021; Wang et al., 2022; Lan-chantin et al., 2023), as well as updating their initial response through self-critique (Press et al.,…”