AI Paper Atlas

Research map / Reinforcement learning / Offline RL

In-Context Compositional Q-Learning for Offline Reinforcement Learning

Qiushui Xu, Yu-Hao Huang, Yushu Jiang, Wenliang Zheng, Lei Song, Jinyu Wang, Jiang Bian

ICLR 2026 · Reinforcement learning › Offline RL · approach: Offline RL sample complexity theory · 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
use ICQL local in-context Q-functions instead of CQL (second best)ICQL · MuJoCo averagenormalized score: 74.0 -> 80.6Improved
use ICQL instead of IQLICQL · Adroit averagenormalized score: 33.2 -> 35.3Improved
use ICQL instead of DT on KitchenICQL · Kitchen averagenormalized score: 55.8 -> 66.8Improved
use ICQL instead of IQL on Hammer-HumanICQL · Hammer-Human-v1normalized score: 7.2 -> 3.7Hurt
use state-similar retrieval instead of random retrieval (ablation)ICQL · Kitchen-Partial-v0normalized score: 47.5 -> 61.5Improved
use state-similar retrieval instead of random retrieval (ablation)ICQL · Hopper-Medium-v2normalized score: 74.1 -> 62.6Hurt
add reward filtering to state-similar retrieval (ablation)ICQL · Walker2d-Medium-v2 / Kitchen-Completenormalized score: 80.3 -> 83.9 (Walker2d-Medium); 79.3 -> 71.3 (Kitchen-Complete)Mixed
use linear Transformer instead of standard Transformer (ablation)ICQL · Walker2d-Medium-Replaynormalized score: 42.9 -> 81.9Improved
use linear Transformer instead of linear MLP (ablation)ICQL · Pen-Humannormalized score: 66.6 -> 85.6Improved
use linear MLP instead of linear Transformer on Hammer (ablation)ICQL · Hammer-Clonednormalized score: 4.5 -> 7.9Improved
increase number of in-context layers from 4 to 20 (ablation)ICQL · MuJoCo tasksnormalized scoreMixed
increase context length beyond 20 (ablation)ICQL · Gym tasksnormalized scoreHurt

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