AI Paper Atlas

Research map / Reinforcement learning / Visual and representation RL

Task Tokens: A Flexible Approach to Adapting Behavior Foundation Models

Ron Vainshtein, Zohar Rimon, Shie Mannor, Chen Tessler

ICLR 2026 · Reinforcement learning › Visual and representation RL · approach: Dexterous manipulation transfer · 4 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
adapt MaskedMimic with Task Tokens instead of full fine-tuningMaskedMimic · Long Jumpsuccess rate: 47.36 -> 99.75Improved
Task Tokens vs MaskedMimic fine-tuning on StrikeMaskedMimic · Strikesuccess rate: 83.07 vs 76.61Hurt
Task Tokens vs fine-tuning on Reach/Direction/SteeringMaskedMimic · Reach/Direction/Steeringsuccess rate: 94.88 vs 93.70; 99.26 vs 99.10; 88.69 vs 87.44No clear effect
add Task Tokens over zero-shot MaskedMimic J.C. onlyMaskedMimic · Reachsuccess rate: 24.77 -> 94.88Improved
use Task Tokens instead of PULSE on SteeringPULSE · Steeringsuccess rate: 40.72 -> 88.69Improved
add joint conditioning to Task Tokens (ablation)Task Tokens · Steeringsuccess rate: 83.66 -> 88.69Improved
add joint conditioning to Task Tokens (ablation)Task Tokens · Directionsuccess rate: 96.89 -> 99.26Improved
add joint conditioning to Task Tokens (ablation)Task Tokens · Reachsuccess rate: 95.37 -> 94.88No clear effect
add joint conditioning to fine-tuning (ablation)MaskedMimic F.T. · Steeringsuccess rate: 87.44 -> 96.41Improved
use current pose input with smaller encoder, no J.C. (ablation)Task Encoder · Steeringsuccess rate: 84.28 -> 66.31Hurt
use bigger MLP encoder with J.C. and current pose (ablation)Task Encoder · Steeringsuccess rate: 79.47 -> 87.77Improved
Task Tokens vs MaskedMimic fine-tuning on human-likenessMaskedMimic · human study, Directionwin rate: 99% ± 1%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