mlsanityData to 13 Oct 2026

Emergent World Models and Latent Variable Estimation in Chess-Playing Language Models

arXiv 2403.15498 ↗First posted 21 Mar 2024cs.LG cs.CLcited by 60

On the map: World models → World models for language agents

Abstract (arXiv)

Language models have shown unprecedented capabilities, sparking debate over the source of their performance. Is it merely the outcome of learning syntactic patterns and surface level statistics, or do they extract semantics and a world model from the text? Prior work by Li et al. investigated this by training a GPT model on synthetic, randomly generated Othello games and found that the model learned an internal representation of the board state. We extend this work into the more complex domain of chess, training on real games and investigating our model's internal representations using linear probes and contrastive activations. The model is given no a priori knowledge of the game and is solely trained on next character prediction, yet we find evidence of internal representations of board state. We validate these internal representations by using them to make interventions on the model's activations and edit its internal board state. Unlike Li et al's prior synthetic dataset approach, our analysis finds that the model also learns to estimate latent variables like player skill to better predict the next character. We derive a player skill vector and add it to the model, improving the model's win rate by up to 2.6 times.

Key facts. Cited by 60 papers in the map (data through 13 Oct 2026). Placed in World models for language agents, one of 11 directions in World models. Builds on 12 of its 18 references that are in the map. First posted on arXiv 21 Mar 2024.

Builds on

12 of 18 references.

Closest ancestors · 1

Same direction, newest first

  1. 2023Emergent Linear Representations in World Models of Self-Supervised Sequence Models2 Sep

Other references · 11

  1. 2016Understanding intermediate layers using linear classifier probescited by 1,313
  2. 2023Representation Engineering: A Top-Down Approach to AI Transparencycited by 1,286
  3. 2023Inference-Time Intervention: Eliciting Truthful Answers from a Language Modelcited by 1,280
  4. 2022Editing Models with Task Arithmeticcited by 1,101
  5. 2023Steering Llama 2 via Contrastive Activation Additioncited by 1,042
  6. 2021Probing Classifiers: Promises, Shortcomings, and Advancescited by 850
  7. 2022Emergent World Representations: Exploring a Sequence Model Trained on a Synthetic Taskcited by 421
  8. 2022Extracting Latent Steering Vectors from Pretrained Language Modelscited by 221
  9. 2017Learning to Generate Reviews and Discovering Sentimentcited by 148
  10. 2021Acquisition of Chess Knowledge in AlphaZerocited by 122
  11. 2023Eight Things to Know about Large Language Modelscited by 73

Built on by

Cited by year (2026 to 13 Oct)

Citing papers per year: 2024: 9; 2025: 23; 2026: 28.

60 papers.

Sort:

By area:

  1. 2024Measuring Progress in Dictionary Learning for Language Model Interpretability with Board Game Modelscited by 51
  2. 2025What Has a Foundation Model Found? Using Inductive Bias to Probe for World Modelscited by 46
  3. 2024Evidence of Learned Look-Ahead in a Chess-Playing Neural Networkcited by 34
  4. 2025(How) Do Language Models Track State?cited by 32
  5. 2024Controlling Large Language Model Agents with Entropic Activation Steeringcited by 29
  6. 2025Mixture of Experts Made Intrinsically Interpretablecited by 24
  7. 2024Unlocking the Future: Exploring Look-Ahead Planning Mechanistic Interpretability in Large Language Modelscited by 21
  8. 2025Improving Reasoning Performance in Large Language Models via Representation Engineeringcited by 21
  9. 2024Transcendence: Generative Models Can Outperform The Experts That Train Themcited by 19
  10. 2024Planning in a recurrent neural network that plays Sokobancited by 17
  11. 2025General agents contain world modelscited by 16
  12. 2024Transformers Use Causal World Models in Maze-Solving Taskscited by 13
38 more
  1. 2025Implicit Search via Discrete Diffusion: A Study on Chesscited by 11
  2. 2025Revisiting the Othello World Model Hypothesiscited by 11
  3. 2024Human-aligned Chess with a Bit of Searchcited by 10
  4. 2025Levels of Analysis for Large Language Modelscited by 10
  5. 2025Too Long, Didn't Model: Decomposing LLM Long-Context Understanding With Novelscited by 9
  6. 2024On Logical Extrapolation for Mazes with Recurrent and Implicit Networkscited by 7
  7. 2025A Unified Definition of Hallucination: It's The World Model, Stupid!cited by 6
  8. 2026Chessformer: A Unified Architecture for Chess Modelingcited by 6
  9. 2026Do Language Models Track Entities Across State Changes?cited by 5
  10. 2025Linear Spatial World Models Emerge in Large Language Modelscited by 4
  11. 2025Path Channels and Plan Extension Kernels: a Mechanistic Description of Planning in a Sokoban RNNcited by 4
  12. 2025Better World Models Can Lead to Better Post-Training Performancecited by 4
  13. 2025Enhancing Conversational Agents with Theory of Mind: Aligning Beliefs, Desires, and Intentions for Human-Like Interactioncited by 3
  14. 2025Beyond World Models: Rethinking Understanding in AI Modelscited by 3
  15. 2026MetaOthello: A Controlled Study of Multiple World Models in Transformerscited by 3
  16. 2025Baba is LLM: Reasoning in a Game with Dynamic Rulescited by 2
  17. 2025From Prediction to Understanding: Will AI Foundation Models Transform Brain Science?cited by 2
  18. 2026Hierarchical Experimentalist Agentscited by 2
  19. 2025Reading Smiles: Proxy Bias in Foundation Models for Facial Emotion Recognitioncited by 1
  20. 2025Emergent World Beliefs: Exploring Transformers in Stochastic Gamescited by 1
  21. 2026Mixture of Masters: Sparse Chess Language Models with Player Routingcited by 1
  22. 2026Verification of the Implicit World Model in a Generative Model via Adversarial Sequencescited by 1
  23. 2026Chess-World-Model: A 10M-Game Benchmark for Exact State Tracking from Chess Move Sequencescited by 1
  24. 2026Latent Programming Horizons in Coding Agentscited by 1
  25. 2025Understanding the learned look-ahead behavior of chess neural networkscited by 0
  26. 2025What if Othello-Playing Language Models Could See?cited by 0
  27. 2025Causal Masking on Spatial Data: An Information-Theoretic Case for Learning Spatial Datasets with Unimodal Language Modelscited by 0
  28. 2025PragWorld: A Benchmark Evaluating LLMs' Local World Model under Minimal Linguistic Alterations and Conversational Dynamicscited by 0
  29. 2026Mechanistic Indicators of Steering Effectiveness in Large Language Modelscited by 0
  30. 2026Understanding the Emergence of Seemingly Useless Features in Next-Token Predictorscited by 0
  31. 2026Predictive Statistics Shape Emergent World Representations of Grid Walkerscited by 0
  32. 2026Tracking vs. Deciding: The Dual-Capability Bottleneck in Searchless Chess Transformerscited by 0
  33. 2026A geometric relation of the error introduced by sampling a language model's output distribution to its internal statecited by 0
  34. 2026Correcting Influence: Unboxing LLM Outputs with Orthogonal Latent Spacescited by 0
  35. 2026Why We Need World Models for AGI: Where LLMs Fail and How World Models May Outperformcited by 0
  36. 2026A Close Look At World Model Recovery In Supervised Fine-Tuned LLM Plannerscited by 0
  37. 2026From Tokens to States: LLMs as a Special Case of World Models and the Continuous Path Beyondcited by 0
  38. 2026One mechanism for many mental spaces: a shared router over a value slot in language modelscited by 0

50 most cited of 60 All 60 (JSON)

Similar, not cited

Close in wording, no citation either way

  1. 2026Disentangling generalization and memorization in large language models using chess
  2. 2023Large Language Models on the Chessboard: A Study on ChatGPT's Formal Language Comprehension and Complex Reasoning Skills
  3. 2024Mastering Board Games by External and Internal Planning with Language Models
  4. 2026How Reasoning Evolves from Post-Training Data: An Empirical Study Using Chess
  5. 2025LLM CHESS: Benchmarking Reasoning and Instruction-Following in LLMs through Chess
  6. 2025ChessArena: A Chess Testbed for Evaluating Strategic Reasoning Capabilities of Large Language Models
  7. 2024Explore the Reasoning Capability of LLMs in the Chess Testbed
  8. 2026Exploring Collaboration between a language and a non-language agent