paper-with-me

Papers

Structured World Representations in Maze-Solving Transformers

2023-12-05 · Michael Igorevich Ivanitskiy, Alex F. Spies, Tilman Räuker, Guillaume Corlouer, Chris Mathwin, Lucia Quirke, Can Rager, Rusheb Shah, Dan Valentine, Cecilia Diniz Behn, Katsumi Inoue, Samy Wu Fung

Transformer models underpin many recent advances in practical machine learning applications, yet understanding their internal behavior continues to elude researchers. Given the size and complexity of these models, forming a comprehensive picture of their inner workings remains a significant challenge. To this end, we set out to understand small transformer models in a more tractable setting: that of solving mazes. In this work, we focus on the abstractions formed by these models and find evidence for the consistent emergence of structured internal representations of maze topology and valid paths. We demonstrate this by showing that the residual stream of only a single token can be linearly decoded to faithfully reconstruct the entire maze. We also find that the learned embeddings of individual tokens have spatial structure. Furthermore, we take steps towards deciphering the circuity of path-following by identifying attention heads (dubbed $\textit{adjacency heads}$), which are implicated in finding valid subsequent tokens.

📄 PDF Abstract BibTeX arXiv:2312.02566

Code (1)

understanding-search/structured-representations-maze-transformers 공식 구현

Tasks

valid

Methods 이 논문이 사용한 방법론

SET Dynamic Sparse Training method where weight mask is updated randomly periodically
Focus 설명 없음

Similar Papers 제목 키워드 기반

Transformers Use Causal World Models in Maze-Solving Tasks

2024-12-16 · Alex F. Spies, William Edwards, Michael I. Ivanitskiy, Adrians Skapars 외

Recent studies in interpretability have explored the inner workings of transformer models trained on tasks across various domains, often discovering that these networks naturally develop surprisingly structured represent…

AGI Maze as a Benchmark Framework for World-Modeling Agents

2026-07-01 · Alexey Potapov arxiv

Large language models (LLMs) are powerful pattern-completion systems, but their default operating mode - predicting the next token from a static context - does not reliably produce persistent, manipulable representations…

Transformers Can Navigate Mazes With Multi-Step Prediction

2024-12-06 · Niklas Nolte, Ouail Kitouni, Adina Williams, Mike Rabbat 외

Despite their remarkable success in language modeling, transformers trained to predict the next token in a sequence struggle with long-term planning. This limitation is particularly evident in tasks requiring foresight t…

GPULanguage ModelingLanguage ModellingNavigate+1

Do LLMs Build Spatial World Models? Evidence from Grid-World Maze Tasks

2026-04-12 · Weijiang Li, Yilin Zhu, Rajarshi Das, Parijat Dube arxiv

Foundation models have shown remarkable performance across diverse tasks, yet their ability to construct internal spatial world models for reasoning and planning remains unclear. We systematically evaluate the spatial un…

Spatial Reasoning

MazeMate: An LLM-Powered Chatbot to Support Computational Thinking in Gamified Programming Learning

2025-09-24 · Chenyu Hou, Hua Yu, Gaoxia Zhu, John Derek Anas 외 arxiv

Computational Thinking (CT) is a foundational problem-solving skill, and gamified programming environments are a widely adopted approach to cultivating it. While large language models (LLMs) provide on-demand programming…