mlsanityData to 7 Oct 2026

Amortized Planning with Large-Scale Transformers: A Case Study on Chess

arXiv 2402.04494 ↗First posted 7 Feb 2024cs.LG cs.AI stat.MLcited by 45

Not on the map. Why?

Abstract (arXiv)

This paper uses chess, a landmark planning problem in AI, to assess transformers' performance on a planning task where memorization is futile unicodex2013 even at a large scale. To this end, we release ChessBench, a large-scale benchmark dataset of 10 million chess games with legal move and value annotations (15 billion data points) provided by Stockfish 16, the state-of-the-art chess engine. We train transformers with up to 270 million parameters on ChessBench via supervised learning and perform extensive ablations to assess the impact of dataset size, model size, architecture type, and different prediction targets (state-values, action-values, and behavioral cloning). Our largest models learn to predict action-values for novel boards quite accurately, implying highly non-trivial generalization. Despite performing no explicit search, our resulting chess policy solves challenging chess puzzles and achieves a surprisingly strong Lichess blitz Elo of 2895 against humans (grandmaster level). We also compare to Leela Chess Zero and AlphaZero (trained without supervision via self-play) with and without search. We show that, although a remarkably good approximation of Stockfish's search-based algorithm can be distilled into large-scale transformers via supervised learning, perfect distillation is still beyond reach, thus making ChessBench well-suited for future research.

Key facts. Cited by 45 papers in the map (data through 7 Oct 2026). Builds on 20 of its 46 references that are in the map. First posted on arXiv 7 Feb 2024.

Builds on

20 of 46 references.

References · 20

  1. 2017Attention Is All You Needcited by 59,932
  2. 2023LLaMA: Open and Efficient Foundation Language Modelscited by 13,958
  3. 2023Llama 2: Open Foundation and Fine-Tuned Chat Modelscited by 12,865
  4. 2020GLU Variants Improve Transformercited by 1,724
  5. 2019Mastering Atari, Go, Chess and Shogi by Planning with a Learned Modelcited by 1,327
  6. 2017Mastering Chess and Shogi by Self-Play with a General Reinforcement Learning Algorithmcited by 723
  7. 2017Convolutional Sequence to Sequence Learningcited by 550
  8. 2019Mish: A Self Regularized Non-Monotonic Activation Functioncited by 141
  9. 2020Aligning Superhuman AI with Human Behavior: Chess as a Model Systemcited by 78
  10. 2021Chess as a Testbed for Language Model State Trackingcited by 69
  11. 2018Improving Regression Performance with Distributional Lossescited by 66
  12. 2020The Chess Transformer: Mastering Play using Generative Language Modelscited by 36
8 more
  1. 2020Learning Models of Individual Behavior in Chesscited by 22
  2. 2017DeepChess: End-to-End Deep Neural Network for Automatic Learning in Chesscited by 12
  3. 2015Giraffe: Using Deep Reinforcement Learning to Play Chesscited by 11
  4. 2019SentiMATE: Learning to play Chess through Natural Language Processingcited by 8
  5. 2022Learning Chess With Language Models and Transformerscited by 6
  6. 2014Move Evaluation in Go Using Deep Convolutional Neural Networkscited by 2
  7. 2023Learning to Play Chess from Textbooks (LEAP): a Corpus for Evaluating Chess Moves based on Sentiment Analysiscited by 2
  8. 2018Deep Pepper: Expert Iteration based Chess agent in the Reinforcement Learning Settingcited by 1

Built on by

Cited by year (2026 to 7 Oct)

Citing papers per year: 2024: 15; 2025: 12; 2026: 18.

45 papers.

Sort:

By area:

  1. 2024Stop Regressing: Training Value Functions via Classification for Scalable Deep RLcited by 136
  2. 2025VideoWorld: Exploring Knowledge Learning from Unlabeled Videoscited by 42
  3. 2024Causal Language Modeling Can Elicit Search and Reasoning Capabilities on Logic Puzzlescited by 35
  4. 2024LMAct: A Benchmark for In-Context Imitation Learning with Long Multimodal Demonstrationscited by 32
  5. 2024Transformers Struggle to Learn to Searchcited by 19
  6. 2024MAPF-GPT: Imitation Learning for Multi-Agent Pathfinding at Scalecited by 17
  7. 2024Explore the Reasoning Capability of LLMs in the Chess Testbedcited by 17
  8. 2024Can Large Language Models Play Games? A Case Study of A Self-Play Approachcited by 12
  9. 2025LLM CHESS: Benchmarking Reasoning and Instruction-Following in LLMs through Chesscited by 12
  10. 2025Graph-Aware Isomorphic Attention for Adaptive Dynamics in Transformerscited by 6
  11. 2026Chessformer: A Unified Architecture for Chess Modelingcited by 6
  12. 2024Lines of Thought in Large Language Modelscited by 5
33 more
  1. 2025Advancing Learnable Multi-Agent Pathfinding Solvers with Active Fine-Tuningcited by 5
  2. 2024BBA: Bi-Modal Behavioral Alignment for Reasoning with Large Vision-Language Modelscited by 4
  3. 2024Chess Rating Estimation from Moves and Clock Times Using a CNN-LSTMcited by 4
  4. 2025Learning to Plan via Supervised Contrastive Learning and Strategic Interpolation: A Chess Case Studycited by 4
  5. 2025ChessArena: A Chess Testbed for Evaluating Strategic Reasoning Capabilities of Large Language Modelscited by 3
  6. 2025Out-of-distribution Tests Reveal Compositionality in Chess Transformerscited by 2
  7. 2025Exploring Human-AI Conceptual Alignment through the Prism of Chesscited by 2
  8. 2026Tracing the Thought of a Grandmaster-level Chess-Playing Transformercited by 2
  9. 2026Understanding Reasoning from Pretraining to Post-Trainingcited by 2
  10. 2024Learning the Latent Rules of a Game from Data: A Chess Storycited by 1
  11. 2025Genomic Next-Token Predictors are In-Context Learnerscited by 1
  12. 2026Disentangling generalization and memorization in large language models using chesscited by 1
  13. 2026Mixture of Masters: Sparse Chess Language Models with Player Routingcited by 1
  14. 2026Learning to Communicate Locally for Large-Scale Multi-Agent Pathfindingcited by 1
  15. 2026GAMBIT: A Three-Mode Benchmark for Adversarial Robustness in Multi-Agent LLM Collectivescited by 1
  16. 2024A proximal policy optimization based intelligent home solar managementcited by 0
  17. 2024AlphaViT: A flexible game-playing AI for multiple games and variable board sizescited by 0
  18. 2024Mastering Chinese Chess AI (Xiangqi) Without Searchcited by 0
  19. 2024Bridging Local and Global Knowledge via Transformer in Board Gamescited by 0
  20. 2025Biological Processing Units: Leveraging an Insect Connectome to Pioneer Biofidelic Neural Architecturescited by 0
  21. 2025Programmatic Representation Learning with Language Modelscited by 0
  22. 2025Causal Masking on Spatial Data: An Information-Theoretic Case for Learning Spatial Datasets with Unimodal Language Modelscited by 0
  23. 2026Tracking vs. Deciding: The Dual-Capability Bottleneck in Searchless Chess Transformerscited by 0
  24. 2026PAWN: Piece Value Analysis with Neural Networkscited by 0
  25. 2026ChessMimic: Per-Rating Transformer Models for Human Move, Clock, and Outcome Prediction in Online Blitz Chesscited by 0
  26. 2026Learning Stateful Predictive Knowledge From Experiencecited by 0
  27. 2026Otter: A Time-Aware, History-Conditioned Human Chess AIcited by 0
  28. 2026PureTD: Reinforcement Learning for Backgammon Money Games with No Evaluation-time Searchcited by 0
  29. 2026Beyond Search-Imitation: Prior-Directed Exploration for Searchless Chesscited by 0
  30. 2026What Counts as Strategic Reasoning? A Systematic Mapping of Chess Research on Humans, Engines, and Language Modelscited by 0
  31. 2026Self-Play Search Distillation for Large Language Model Reasoningcited by 0
  32. 2026Conditional Generation of Creative Chess Puzzles with Diffusion Modelscited by 0
  33. 2026Bongard: Training Machine Intuitioncited by 0

Similar, not cited

Close in wording, no citation either way

  1. 2026Chess-World-Model: A 10M-Game Benchmark for Exact State Tracking from Chess Move Sequences
  2. 2024Mastering Board Games by External and Internal Planning with Language Models
  3. 2025Xiangqi-R1: Enhancing Spatial Strategic Reasoning in LLMs for Chinese Chess via Reinforcement Learning
  4. 2025Implicit Search via Discrete Diffusion: A Study on Chess
  5. 2024Checkmating One, by Using Many: Combining Mixture of Experts with MCTS to Improve in Chess
  6. 2026How Reasoning Evolves from Post-Training Data: An Empirical Study Using Chess
  7. 2026Finding the Move Is Not Winning the Game: XiangqiBench for Closed-Loop Evaluation of LLM Agents
  8. 2026Oracle-Guided Soft Shielding for Safe Move Prediction in Chess