mlsanityData to 7 Oct 2026

Structure and Scale in Simplicial Sequence Modelling

arXiv 2606.01302 ↗First posted 31 May 2026cs.LGcited by 0

Not on the map. Why?

Abstract (arXiv)

Modern large-scale deep learning exhibits two striking empirical phenomena: behavioural scaling laws (predictable performance gains with increasing scale) and emergent mechanisms (structured internal representations and circuits in deep neural networks). We hypothesise that these two phenomena are connected: that predictable changes in behaviour are the result of predictable changes in internal computational structure. In this paper, we report preliminary evidence of such a connection. We find a correlation between scaling patterns in performance and representations in small transformers trained to predict the outputs of a hidden Markov model, for which residual activations are known to linearly encode a belief distribution over latent states in a probability simplex.

Key facts. No citing papers in the map yet (data through 7 Oct 2026). Builds on 48 of its 64 references that are in the map. First posted on arXiv 31 May 2026.

Builds on

48 of 64 references.

References · 48

  1. 2017Attention Is All You Needcited by 59,932
  2. 2015Deep Residual Learning for Image Recognitioncited by 51,809
  3. 2020Language Models are Few-Shot Learnerscited by 32,199
  4. 2020Scaling Laws for Neural Language Modelscited by 6,411
  5. 2014Neural Machine Translation by Jointly Learning to Align and Translatecited by 4,530
  6. 2022Training Compute-Optimal Large Language Modelscited by 2,927
  7. 2016Understanding intermediate layers using linear classifier probescited by 1,279
  8. 2022Interpretability in the Wild: a Circuit for Indirect Object Identification in GPT-2 smallcited by 1,130
  9. 2022In-context Learning and Induction Headscited by 930
  10. 2022Toy Models of Superpositioncited by 928
  11. 2023Progress measures for grokking via mechanistic interpretabilitycited by 822
  12. 2022Grokking: Generalization Beyond Overfitting on Small Algorithmic Datasetscited by 599
36 more
  1. 2017Deep Learning Scaling is Predictable, Empiricallycited by 556
  2. 2019Leveraging Procedural Generation to Benchmark Reinforcement Learningcited by 478
  3. 2022Emergent World Representations: Exploring a Sequence Model Trained on a Synthetic Taskcited by 417
  4. 2018Quantifying Generalization in Reinforcement Learningcited by 345
  5. 2023Emergent Linear Representations in World Models of Self-Supervised Sequence Modelscited by 313
  6. 2019A Constructive Prediction of the Generalization Error Across Scalescited by 202
  7. 2024Not All Language Model Features Are One-Dimensionally Linearcited by 201
  8. 2015Visualizing and Understanding Recurrent Networkscited by 158
  9. 2023The Clock and the Pizza: Two Stories in Mechanistic Explanation of Neural Networkscited by 153
  10. 2017Learning to Generate Reviews and Discovering Sentimentcited by 147
  11. 2023Pretraining task diversity and the emergence of non-Bayesian in-context learning for regressioncited by 138
  12. 2023The Quantization Model of Neural Scalingcited by 137
  13. 2023Sudden Drops in the Loss: Syntax Acquisition, Phase Transitions, and Simplicity Bias in MLMscited by 121
  14. 2024The Evolution of Statistical Induction Heads: In-Context Learning Markov Chainscited by 115
  15. 2021Learning Curve Theorycited by 79
  16. 2023In-Context Learning through the Bayesian Prismcited by 78
  17. 2024Transformers represent belief state geometry in their residual streamcited by 73
  18. 2023The Transient Nature of Emergent In-Context Learning in Transformerscited by 72
  19. 2024Chinchilla Scaling: A replication attemptcited by 65
  20. 2024Competition Dynamics Shape Algorithmic Phases of In-Context Learningcited by 52
  21. 2020A Neural Scaling Law from the Dimension of the Data Manifoldcited by 48
  22. 2026When Models Manipulate Manifolds: The Geometry of a Counting Taskcited by 43
  23. 2024Loss Landscape Degeneracy and Stagewise Development in Transformerscited by 39
  24. 2024Differentiation and Specialization of Attention Heads via the Refined Local Learning Coefficientcited by 39
  25. 2023Dynamical versus Bayesian Phase Transitions in a Toy Model of Superpositioncited by 28
  26. 2024Planning in a recurrent neural network that plays Sokobancited by 17
  27. 2025You Are What You Eat -- AI Alignment Requires Understanding How Data Shapes Structure and Generalisationcited by 15
  28. 2025Interpreting Emergent Planning in Model-Free Reinforcement Learningcited by 15
  29. 2025Constrained belief updates explain geometric structures in transformer representationscited by 14
  30. 2025Dynamics of Transient Structure in In-Context Linear Regression Transformerscited by 12
  31. 2024Sometimes I am a Tree: Data Drives Unstable Hierarchical Generalizationcited by 10
  32. 2017Nearly Maximally Predictive Features and Their Dimensionscited by 9
  33. 2025Neural networks leverage nominally quantum and post-quantum representationscited by 5
  34. 2025Path Channels and Plan Extension Kernels: a Mechanistic Description of Planning in a Sokoban RNNcited by 4
  35. 2026Stagewise Reinforcement Learning and the Geometry of the Regret Landscapecited by 3
  36. 2026Time Travel Engine: A Shared Latent Chronological Manifold Enables Historical Navigation in Large Language Modelscited by 2

Built on by

Not cited yet.