SOURCE-LINKED INTELLIGENCE

Few-Shot Demonstrations Elicit the Use of In-Context World Representations in LLMs

arXiv · AI, language, vision and robotics · article · Sep 21, 2026 · UTC

Large language models (LLMs), when acting as agents, are expected to take observed data in context, infer the latent state space underlying the world, and leverage it for downstream prediction. However, prior work demonstrated that LLMs struggle to use representations learned in context on a graph tracking task, where the model needs to construct a representation of the graph governing data generation process and use it for subsequent predictions. In this paper, we show that extending this to few-shot settings, where each demonstration is generated from a different world with either the same o

Read original source ↗ Open in workspace

recordType

paper

region

Global