AI Paper Atlas

Research map / Multi-agent systems and games / Human-AI and goal-conditioned

Collaborating with Humans without Human Data

DJ Strouse, Kevin R. McKee, Matthew Botvinick, Edward Hughes, Richard Everett

NeurIPS 2021 · Multi-agent systems and games › Human-AI and goal-conditioned · approach: Human-AI coordination agents · 3 citations (OpenAlex)

Read the paperCheck your idea against 137,217 papers

AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
add past partner checkpoints representing different training skill levels (ablation)FCP · H_proxydeliveries: 4.7±0.4 -> 10.6±0.5Improved
add past partner checkpoints representing different training skill levels (ablation)FCP · Diverse SPdeliveries: 6.9±0.1 -> 11.2±0.1Improved
add past partner checkpoints representing different training skill levels (ablation)FCP · Randomdeliveries: 1.0±0.1 -> 8.6±0.2Improved
add architectural variation to the partner population (ablation)FCP with past checkpoints · H_proxydeliveries: 10.6±0.5 -> 9.9±0.6No clear effect
add architectural variation to the partner population (ablation)FCP with past checkpoints · Diverse SPdeliveries: 11.2±0.1 -> 11.1±0.4No clear effect
add architectural variation to the partner population (ablation)FCP with past checkpoints · Randomdeliveries: 8.6±0.2 -> 8.4±0.4No clear effect
add architectural variation when past checkpoints are unavailable (ablation)FCP−T · H_proxydeliveries: 4.7±0.4 -> 7.0±0.8Improved
add architectural variation when past checkpoints are unavailable (ablation)FCP−T · Diverse SPdeliveries: 6.9±0.1 -> 8.6±0.4Improved
add architectural variation when past checkpoints are unavailable (ablation)FCP−T · Randomdeliveries: 1.0±0.1 -> 3.2±0.5Improved
increase partner population size from N=4 to N=32 (ablation)FCP · H_proxydeliveries: 8.4 (0.3) -> 10.6 (0.5)Improved
increase partner population size from N=32 to N=64 (ablation)FCP · H_proxydeliveries: 10.6 (0.5) -> 10.4 (0.3)No clear effect
train with diverse self-play partners and their past checkpoints instead of self-play, population-play, or behavioral-cloning playFCP · held-out agent partners and human partnersdeliveriesImproved

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy