AI Paper Atlas

Research map / Reinforcement learning / Visual and representation RL

Off-Team Learning

Brandon Cui, Hengyuan Hu, Andrei Lupu, Samuel Sokota, Jakob Nicolaus Foerster

NeurIPS 2022 · Reinforcement learning › Visual and representation RL · approach: Probabilistic and autonomous RL · 0 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
Train belief model off-team (OT-BL) instead of vanilla belief learningOBL · 2-life Hanabi, level 1, with invalid card penaltycross-play score: 16.71 ± 0.06 -> 21.15 ± 0.02Improved
OT-BL instead of vanilla belief learningOBL · 2-life Hanabi, level 3, no penaltycross-play score: 23.05 ± 0.30 -> 23.76 ± 0.02Improved
OT-BL instead of vanilla belief learningOBL · 2-life Hanabi, level 1, no penaltySXG: 0.84 -> 0.05Improved
OT-BL instead of vanilla belief learningbelief model · 2-life Hanabi with penalty, level 1 policy training% valid fictitious samples: ~83% -> > 99.5%Improved
Add invalid card penalty to environmentvanilla OBL level 1 · 2-life Hanabicross-play score: 20.17 ± 0.15 -> 16.71 ± 0.06Hurt
Add invalid card penalty to environmentOT-BL level 1 · 2-life Hanabicross-play score: 21.05 ± 0.02 -> 21.15 ± 0.02No clear effect
OT-OBL instead of vanilla OBLOBL · 3-life Hanabi, level 5cross-play score: 23.99 ± 0.11 -> 24.24 ± 0.01Improved
OT-OBL at level 1OBL · 3-life Hanabi, level 1cross-play score: 21.50 ± 0.05 -> 21.31 ± 0.01Hurt
Add robust belief (belief trained on broader distributions) to OT-OBLOT-OBL · 3-life Hanabi, level 5cross-play score: 24.24 ± 0.01 -> 24.30 ± 0.01Improved
OT-OBL with 3200 rank bot trajectoriesOBL · ad-hoc teamplay with rank botscore: 14.47 ± 0.67 -> 15.66 ± 0.56Improved
OT-OBL with 3200 clone bot trajectoriesOBL · proxy human-AI coordinationscore: 16.19 ± 0.21 -> 16.64 ± 0.20Improved
OT-OBL with clone bot trajectories evaluated against rank botOBL · ad-hoc teamplay with rank botscore: 14.47 ± 0.67 -> 14.33 ± 0.63No clear effect

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