AI Paper Atlas

Research map / Large language models / Language, safety and interpretability

Context and Diversity Matter: The Emergence of In-Context Learning in World Models

Fan Wang, ZHIYUAN CHEN, YUXUAN ZHONG, Sunjian Zheng, Pengtao Shao, Bo Yu, Shaoshan Liu, Jianan Wang, Ning Ding, Yang Cao, Yu Kang

ICLR 2026 · Large language models › Language, safety and interpretability · approach: In-context learning analysis

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
increase context length from T=1 to T=1000L2World (Maze-32K-L) · Maze unseen envsPSNR: 16.37 -> 24.66Improved
train on 32K diverse environments instead of 128 (ablation)L2World · Maze unseen envs, T=10000PSNR: 21.52 -> 24.65Improved
train on 128 environments instead of 32K (ablation)L2World · Maze seen envs, T=10000PSNR: 25.05 -> 26.00Improved
use long trajectories (10K) instead of short (100) (ablation)L2World, 32K envs · Maze unseen envs, T=10000PSNR: 20.31 -> 24.65Improved
use L2World instead of Dreamer (LSTM)L2World (Maze-32K-L) · Maze unseen envs, T=10000PSNR: 22.12 -> 24.65Improved
use L2World instead of NWM (diffusion, 4 frames)L2World (Maze-32K-L) · Maze unseen envs, T=10000PSNR: 17.85 -> 24.65Improved
pre-train on Maze-32K-L before ProcTHOR-5K (ablation)L2World · ProcTHOR-5K unseen, T=10000PSNR: 19.81 -> 22.04Improved
increase ProcTHOR data from 5K to 40K (2K-length trajectories) (ablation)L2World pre-trained on Maze-32K-L · ProcTHOR unseen, T=10000PSNR: 22.04 -> 23.94Mixed
train on 4 environments instead of many (cart-pole) (ablation)world model · Random cart-pole unseen envsprediction errorHurt
train on both Scope 1+2 with 8K envs instead of Scope 1 only (ablation)world model · Random cart-poleprediction errorImproved
early-stop checkpoint instead of over-training (4 envs) (ablation)world model · Random cart-pole unseen envsprediction errorImproved
shuffle 20% or 50% of context observations (ablation)L2World (Maze-32K-L vs Maze-128-L) · MazePSNRHurt

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