Phase Transitions in Attention: A Bayesian Theory of Copy Head Emergence

• URL: http://arxiv.org/abs/2606.12058v1
• Date: Wed, 10 Jun 2026 13:26:56 GMT
• Title: Phase Transitions in Attention: A Bayesian Theory of Copy Head Emergence
• Abstract summary: We show how the copy subcircuit in the first layer of an induction head is learned by analyzing a single-layer softmax attention network trained on a copy task.<n>Our work provides a first-principles theoretical account of the abrupt emergence of the copy subcircuit, reminiscent of the one observed in training large language models.
• Score: 6.025438902954768
• License: http://creativecommons.org/licenses/by/4.0/
• Abstract: Attention is the key mechanism underlying in-context learning in transformers, and attention patterns have been observed empirically to emerge abruptly during training. We present a Bayesian theory of feature learning in attention; we then focus on how the copy subcircuit in the first layer of an induction head is learned by analyzing a single-layer softmax attention network trained on a copy task. We derive a closed-form posterior over the attention matrix and reduce it to a low-dimensional order parameter space. This reduction reveals a phase transition in the amount of training data, which we verify using both Bayesian sampling and standard training with Adam. We contrast our results with linear attention and find that softmax attention exhibits a \emph{first-order phase transition} while in linear attention an initial \emph{second-order phase transition} is followed by a smooth, continuous evolution toward the structured attention pattern (\emph{crossover}). Our work provides a first-principles theoretical account of the abrupt emergence of the copy subcircuit, reminiscent of the one observed in training large language models.

Related papers

• Focus and Dilution: The Multi-stage Learning Process of Attention [9.240393019828968]
  We show that a single focus-dilution cycle can be decomposed into a sequence of distinct stages.<n>Experiments on synthetic Markovian data as well as WikiText and TinyStories corroborate the predicted stages and cyclical dynamics.
  arXiv  Detail & Related papers  (2026-05-02T02:30:52Z)
• From Density Matrices to Phase Transitions in Deep Learning: Spectral Early Warnings and Interpretability [0.0]
  We show that this object provides a computationally efficient, unified observable of phase transitions during training.<n>Remarkably, the top eigenvectors of the 2RDM are directly interpretable making it straightforward to study the nature of the transitions.
  arXiv  Detail & Related papers  (2026-03-31T14:34:01Z)
• Specialization of softmax attention heads: insights from the high-dimensional single-location model [0.0]
  We study the training dynamics of multi-head softmax attention under SGD.<n>We show that softmax-1 significantly reduces noise from irrelevant heads.<n>We introduce the Bayes-softmax attention, which achieves optimal prediction performance in this setting.
  arXiv  Detail & Related papers  (2026-03-04T12:37:08Z)
• How Attention Sinks Emerge in Large Language Models: An Interpretability Perspective [67.08306259095778]
  We identify a simple mechanism that enables the model to recognize token at position zero and induce an attention sink within two transformer blocks.<n>By analyzing training traces from a 30B A3B MoE model trained from scratch, we find that this mechanism emerges early in training and becomes increasingly concentrated in the first two layers.
  arXiv  Detail & Related papers  (2026-02-04T11:10:34Z)
• How Transformers Learn Regular Language Recognition: A Theoretical Study on Training Dynamics and Implicit Bias [48.9399496805422]
  We focus on two representative tasks in the category of regular language recognition, known as even pairs' and parity check'<n>Our goal is to explore how a one-layer transformer, consisting of an attention layer followed by a linear layer, learns to solve these tasks.
  arXiv  Detail & Related papers  (2025-05-02T00:07:35Z)
• In-Context Linear Regression Demystified: Training Dynamics and Mechanistic Interpretability of Multi-Head Softmax Attention [52.159541540613915]
  We study how multi-head softmax attention models are trained to perform in-context learning on linear data.<n>Our results reveal that in-context learning ability emerges from the trained transformer as an aggregated effect of its architecture and the underlying data distribution.
  arXiv  Detail & Related papers  (2025-03-17T02:00:49Z)
• On the Emergence of Position Bias in Transformers [59.87743433861665]
  This paper presents a graph-theoretic framework for analyzing position biases in multilayer positions.<n>Our framework offers a principled foundation for understanding positional interplay in transformers.
  arXiv  Detail & Related papers  (2025-02-04T02:53:07Z)
• Training Dynamics of Transformers to Recognize Word Co-occurrence via Gradient Flow Analysis [97.54180451650122]
  We study the dynamics of training a shallow transformer on a task of recognizing co-occurrence of two designated words. We analyze the gradient flow dynamics of simultaneously training three attention matrices and a linear layer. We prove a novel property of the gradient flow, termed textitautomatic balancing of gradients, which enables the loss values of different samples to decrease almost at the same rate and further facilitates the proof of near minimum training loss.
  arXiv  Detail & Related papers  (2024-10-12T17:50:58Z)
• Training Dynamics of Multi-Head Softmax Attention for In-Context Learning: Emergence, Convergence, and Optimality [54.20763128054692]
  We study the dynamics of gradient flow for training a multi-head softmax attention model for in-context learning of multi-task linear regression. We prove that an interesting "task allocation" phenomenon emerges during the gradient flow dynamics.
  arXiv  Detail & Related papers  (2024-02-29T18:43:52Z)
• A phase transition between positional and semantic learning in a solvable model of dot-product attention [30.96921029675713]
  Morelinear model dot-product attention is studied as a non-dimensional self-attention layer with trainable and low-dimensional query and key data. We show that either a positional attention mechanism (with tokens each other based on their respective positions) or a semantic attention mechanism (with tokens tied to each other based their meaning) or a transition from the former to the latter with increasing sample complexity.
  arXiv  Detail & Related papers  (2024-02-06T11:13:54Z)
• Exact Phase Transitions in Deep Learning [5.33024001730262]
  We prove that the competition between prediction error and model complexity in the training loss leads to the second-order phase transition for nets with one hidden layer and the first-order phase transition for nets with more than one hidden layer. The proposed theory is directly relevant to the optimization of neural networks and points to an origin of the posterior collapse problem in Bayesian deep learning.
  arXiv  Detail & Related papers  (2022-05-25T06:00:34Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.