AI Paper Atlas

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

Continual Learning for Instruction Following from Realtime Feedback

Alane Suhr, Yoav Artzi

NeurIPS 2023 · Multi-agent systems and games › Human-AI and goal-conditioned · approach: Boundedly rational agent modeling · 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
Train continually on realtime user feedbackInstruction-following agent · CerealBar, long-term experimentinstruction execution accuracy: 66.7 ± 1.5 -> 82.1 ± 1.1Improved
Deploy the continually trained final agent rather than the initial agentInstruction-following agent · CerealBar, side-by-side comparisoninstruction execution accuracy: 70.4 -> 83.6Improved
Add heuristically propagated rewards instead of using only simple rewards (ablation)RewardProp · CerealBar, Round 3instruction execution accuracy: 78.1 ± 1.0 -> 80.3 ± 1.0Improved
Discard negative feedback and reboots from training (ablation)RewardProp · CerealBar, Round 4instruction execution accuracy: 81.0 ± 0.9 -> 76.8 ± 1.0Hurt
Reduce initialization demonstrations to 144 human-human interactions (ablation)Instruction-following agent · CerealBar, initializationinstruction execution accuracy: 70.6 ± 1.1 -> 34.8 ± 1.8Hurt
Reduce initialization demonstrations to 144 human-human interactions (ablation)Instruction-following agent · CerealBar, after five roundsinstruction execution accuracy: 34.8 ± 1.8 -> 64.2 ± 1.5Improved
Use feedback data from human-agent interaction rather than an equivalent amount of added human demonstrations (ablation)RewardProp · CerealBar, five-round comparisoninstruction execution accuracyNo clear effect
Propagate feedback to previously unrewarded actions within the heuristic window (ablation)RewardProp · CerealBar, action reward assignmentproportion of actions assigned reward: 63.1% -> 82.2%Improved

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