[Reading] BlenderBot 3: A deployed conversational agent

This is a new paper from Facebook on August 5, 2022
Abstract:
We present BlenderBot 3, a 175B parameter dialogue model capable of open-domain conversation with access to the internet and a long-term memory, and having been trained on a large number of user defined tasks. We release both the model weights and code, and have also deployed the model on a public web page to interact with organic users. This technical report describes how the model was built (architecture, model and training scheme), and details of its deployment, including safety mechanisms. Human evaluations show its superiority to existing open-domain dialogue agents, including its predecessors (Roller et al., 2021; Komeili et al., 2022). Finally, we detail our plan for continual learning using the data collected from deployment, which will also be publicly released. The goal of this research program is thus to enable the community to study ever-improving responsible agents that learn through interaction.

Paper to read:
Shuster, Kurt, Jing Xu, Mojtaba Komeili, Da Ju, Eric Michael Smith, Stephen Roller, Megan Ung et al. "BlenderBot 3: a deployed conversational agent that continually learns to responsibly engage." arXiv preprint arXiv:2208.03188 (August 5, 2022). https://arxiv.org/abs/2208.03188
Facebook AI Blog, June 30, 2022, August 9, 2022, “BlenderBot 3: A 175B parameter, publicly available chatbot that improves its skills and safety over time”,
Demo: https://blenderbot.ai/
Project website: https://parl.ai/projects/bb3/

Discussion leader: Junling Hu

This is part of the bi-weekly reading series. We come together to discuss cutting-edge AI topics and papers. The discussion is led by one presenter, with group participation.

Join us online at: https://us02web.zoom.us/meeting/register/tZIudeqvqDItGNSHspgNiD-cr2nBKhnWWnHA

7-7:15pm Meet and greet
7:15-8:15pm Paper presentation and group discussion
8:15-8:30 Additional discussions