AI Paper Atlas

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

Discovering Diverse Behaviors via Temporal Contrastive Learning

Faisal Mohamed, Catherine Ji, Benjamin Eysenbach, Glen Berseth

ICLR 2026 · Multi-agent systems and games › Human-AI and goal-conditioned · approach: Intrinsically motivated exploration

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 representation normalization (ablation)C-TeC · Ant, Humanoid, Arm, Craftax-Classicnumber of visited states / achievementsImproved
use a separable critic instead of a monolithic critic (ablation)C-TeC · Ant, Humanoid, Armnumber of visited states / achievementsImproved
use reverse KL (mode-seeking) reward instead of forward KL (ablation)C-TeC · Ant, Humanoid, Armnumber of visited states / achievementsImproved
use InfoNCE instead of binary NCE loss (ablation)C-TeC · Craftax-Classicnumber of visited states / achievementsImproved
swap InfoNCE for symmetric InfoNCE, FlatNCE or FB loss (ablation)C-TeC · Ant, Humanoid, Arm, Craftax-Classicnumber of visited states / achievementsNo clear effect
use L1 distance critic instead of L2 or dot product (ablation)C-TeC · robotic environmentsnumber of visited states / achievementsImproved
use L2 distance critic instead of L1 (ablation)C-TeC · Craftax-Classicnumber of visited states / achievementsImproved
replace geometric future-state sampling with uniform or gamma-schedule sampling (ablation)C-TeC · Ant, Humanoid, Armnumber of visited states / achievementsNo clear effect
use contrastive discount 0.99 versus smaller values (ablation)C-TeC · Ant, Humanoid, Armnumber of visited states / achievementsMixed
fix temperature tau to 1 instead of other fixed values (ablation)C-TeC · Ant, Humanoid, Armnumber of visited states / achievementsImproved
sample in-episode negatives (repetition factor 2 or 4) (ablation)C-TeC · Humanoid-u-maze, Arm-binpick-hardnumber of visited states / achievementsImproved
increase training from 30M to 500M environment steps (ablation)C-TeC · Ant-hardest-mazenumber of visited states / achievements: 1119 -> 2500Improved

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