Publication

Recipes for Safety in Open-domain Chatbots

arXiv


Abstract

Models trained on large unlabeled corpora of human interactions will learn patterns and mimic behaviors therein, which include offensive or otherwise toxic behavior and unwanted biases. We investigate a variety of methods to mitigate these issues in the context of open-domain generative dialogue models. We introduce a new human-and-model-in-the-loop framework for both training safer models and for evaluating them, as well as a novel method to distill safety considerations inside generative models without the use of an external classifier at deployment time. We conduct experiments comparing these methods and find our new techniques are (i) safer than existing models as measured by automatic and human evaluations while (ii) maintaining usability metrics such as engagingness relative to the state of the art. We then discuss the limitations of this work by analyzing failure cases of our models.

Related Publications

All Publications

NeurIPS - December 5, 2021

Interpretable agent communication from scratch (with a generic visual processor emerging on the side)

Roberto Dessì, Eugene Kharitonov, Marco Baroni

Workshop on Online Abuse and Harms (WHOAH) at ACL - November 30, 2021

Findings of the WOAH 5 Shared Task on Fine Grained Hateful Memes Detection

Lambert Mathias, Shaoliang Nie, Bertie Vidgen, Aida Davani, Zeerak Waseem, Douwe Kiela, Vinodkumar Prabhakaran

Journal of Big Data - November 6, 2021

A graphical method of cumulative differences between two subpopulations

Mark Tygert

Electronics (MDPI) Journal - November 4, 2021

Performance Evaluation of Offline Speech Recognition on Edge Devices

Santosh Gondi, Vineel Pratap

To help personalize content, tailor and measure ads, and provide a safer experience, we use cookies. By clicking or navigating the site, you agree to allow our collection of information on and off Facebook through cookies. Learn more, including about available controls: Cookie Policy