Emergent Capabilities Arise Randomly from Learning Sparse Attention Patterns

• URL: http://arxiv.org/abs/2606.25010v1
• Date: Tue, 23 Jun 2026 17:51:10 GMT
• Title: Emergent Capabilities Arise Randomly from Learning Sparse Attention Patterns
• Abstract summary: downstream capabilities such as in-context learning are known to emerge abruptly past a certain model scale.<n>We demonstrate that the emergence of capabilities such as pattern completion and indirect object identification corresponds to the abrupt learning of task-relevant attention patterns.<n>We train transformer models on synthetic linear map and cellular automata datasets, and we show that the difficulty of learning attention patterns depends on context length and pattern sparsity.
• Score: 48.062049009807446
• License: http://creativecommons.org/licenses/by/4.0/
• Abstract: Neural scaling laws for transformer language models predict smooth improvements in pretraining loss with increasing parameters, but downstream capabilities such as in-context learning are known to emerge abruptly past a certain model scale. In this paper, we show that emergent capabilities arise stochastically throughout training, with larger models acquiring them earlier on average. We demonstrate that the emergence of capabilities such as pattern completion and indirect object identification corresponds to the abrupt learning of task-relevant attention patterns. To isolate this phenomenon, we train transformer models on synthetic linear map and cellular automata datasets, and we show that the difficulty of learning attention patterns depends on context length and pattern sparsity. Moreover, scaling the number of attention heads improves learning efficiency on our synthetic tasks, while increasing the head dimension yields diminishing returns past a minimum capacity. We additionally investigate architectures with alternative attention mechanisms, showing that MLP-Mixer outperforms a transformer on linear map tasks with complex attention patterns. Our findings provide a mechanistic insight into emergence, showing that downstream capabilities arise abruptly due to the intrinsic difficulty of learning sparse attention patterns in transformer models.

Related papers

• Transformer Is Inherently a Causal Learner [27.79148022495734]
  We show that transformer trained in an autoregressive manner naturally encodes time-delayed causal structures.<n>We prove this connection theoretically under standard identifiability conditions.<n>This approach greatly surpasses the performance of state-of-the-art discovery algorithms.
  arXiv  Detail & Related papers  (2026-01-09T09:10:04Z)
• Revisiting associative recall in modern recurrent models [17.2249234816671]
  We show that the choice of learning rate plays a critical role in the performance of modern recurrent models.<n>We then inspect 1-layer transformers, revealing that despite their poor performance, their training dynamics surprisingly resemble the formation of induction heads.
  arXiv  Detail & Related papers  (2025-08-26T13:45:08Z)
• The emergence of sparse attention: impact of data distribution and benefits of repetition [14.652502263025882]
  We study the emergence over training of sparse attention, a critical and frequently observed attention pattern in Transformers.<n>By combining theoretical analysis of a toy model with empirical observations on small Transformers trained on a linear regression variant, we uncover the mechanics sparse attention emergence.<n>Our findings provide a simple, theoretically grounded framework for understanding how data distributions and model design influence the learning dynamics behind one form of emergence.
  arXiv  Detail & Related papers  (2025-05-23T13:14:02Z)
• 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)
• Abrupt Learning in Transformers: A Case Study on Matrix Completion [15.210510215283882]
  We formulate the low-rank matrix completion problem as a masked language modeling (MLM) task. We show that it is possible to train a BERT model to solve this task to low error. We also analyze the training dynamics of individual model components to understand the sudden drop in loss.
  arXiv  Detail & Related papers  (2024-10-29T17:08:06Z)
• Localized Gaussians as Self-Attention Weights for Point Clouds Correspondence [92.07601770031236]
  We investigate semantically meaningful patterns in the attention heads of an encoder-only Transformer architecture.<n>We find that fixing the attention weights not only accelerates the training process but also enhances the stability of the optimization.
  arXiv  Detail & Related papers  (2024-09-20T07:41:47Z)
• Research on Personalized Compression Algorithm for Pre-trained Models Based on Homomorphic Entropy Increase [2.6513322539118582]
  We explore the challenges and evolution of two key technologies in the current field of AI: Vision Transformer model and Large Language Model (LLM) Vision Transformer captures global information by splitting images into small pieces, but its high reference count and compute overhead limit deployment on mobile devices. LLM has revolutionized natural language processing, but it also faces huge deployment challenges.
  arXiv  Detail & Related papers  (2024-08-16T11:56:49Z)
• In-Context Convergence of Transformers [63.04956160537308]
  We study the learning dynamics of a one-layer transformer with softmax attention trained via gradient descent. For data with imbalanced features, we show that the learning dynamics take a stage-wise convergence process.
  arXiv  Detail & Related papers  (2023-10-08T17:55:33Z)
• Uncovering the Hidden Cost of Model Compression [43.62624133952414]
  Visual Prompting has emerged as a pivotal method for transfer learning in computer vision. Model compression detrimentally impacts the performance of visual prompting-based transfer. However, negative effects on calibration are not present when models are compressed via quantization.
  arXiv  Detail & Related papers  (2023-08-29T01:47:49Z)
• Stabilizing Transformer Training by Preventing Attention Entropy Collapse [56.45313891694746]
  We investigate the training dynamics of Transformers by examining the evolution of the attention layers. We show that $sigma$Reparam successfully prevents entropy collapse in the attention layers, promoting more stable training. We conduct experiments with $sigma$Reparam on image classification, image self-supervised learning, machine translation, speech recognition, and language modeling tasks.
  arXiv  Detail & Related papers  (2023-03-11T03:30:47Z)
• Data-driven emergence of convolutional structure in neural networks [83.4920717252233]
  We show how fully-connected neural networks solving a discrimination task can learn a convolutional structure directly from their inputs. By carefully designing data models, we show that the emergence of this pattern is triggered by the non-Gaussian, higher-order local structure of the inputs.
  arXiv  Detail & Related papers  (2022-02-01T17:11:13Z)
• SparseBERT: Rethinking the Importance Analysis in Self-attention [107.68072039537311]
  Transformer-based models are popular for natural language processing (NLP) tasks due to its powerful capacity. Attention map visualization of a pre-trained model is one direct method for understanding self-attention mechanism. We propose a Differentiable Attention Mask (DAM) algorithm, which can be also applied in guidance of SparseBERT design.
  arXiv  Detail & Related papers  (2021-02-25T14:13:44Z)

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.