| |
|
| |
|
Join the Snorkel AI Reading Group, a recurring forum to explore the latest frontier developments in AI while building meaningful connections within the community.nnIn this afternoon session, Yijia Shao, a PhD candidate at Stanford NLP, will cover work she recently presented at ICLR: Collaborative Gym: A Framework for Enabling & Evaluating Human-Agent Collaboration. nnAgenda:n3pm - doors openn3:30pm - talk beginsn Boba tea & other refreshments will be provided ! nAmong other things, you'll learn:nHow Co-Gym formalizes humanagent collaboration as a POMDP with dual control over a shared workspace.nWhy enforced turn-taking breaks down for real tasks, & what replaces it: two collaboration acts plus a notification protocol.nHow an evaluation suite scores both outcomes & process, including an entropy-based measure of shared initiative.nWhy a Collaborative Agent with Situational Planning beats fully autonomous baselines across three tasks.nHow real users rate the best collaborative agent: 86% 74% 66% win rates over autonomous on Travel, Tabular, & Related Work.nWhy communication & situational awareness remain the dominant failure modes (65% & 40% of real trajectories).
|
|
|
|
|
|
|
|
