Mechanism and Emergence of Stacked Attention Heads in Multi-Layer Transformers

• URL: http://arxiv.org/abs/2411.12118v2
• Date: Fri, 14 Feb 2025 12:56:45 GMT
• Title: Mechanism and Emergence of Stacked Attention Heads in Multi-Layer Transformers
• Authors: Tiberiu Musat,
• Abstract summary: I introduce the retrieval problem, a simple yet common reasoning task that can be solved only by transformers with a minimum number of layers. I empirically show that large language models can solve the task under different prompting formulations without any fine-tuning.
• Score: 0.0
• License:
• Abstract: In this paper, I introduce the retrieval problem, a simple yet common reasoning task that can be solved only by transformers with a minimum number of layers, which grows logarithmically with the input size. I empirically show that large language models can solve the task under different prompting formulations without any fine-tuning. To understand how transformers solve the retrieval problem, I train several transformers on a minimal formulation. Successful learning occurs only under the presence of an implicit curriculum. I uncover the learned mechanisms by studying the attention maps in the trained transformers. I also study the training process, uncovering that attention heads always emerge in a specific sequence guided by the implicit curriculum.

Related papers

• One-Layer Transformer Provably Learns One-Nearest Neighbor In Context [48.4979348643494]
  We study the capability of one-layer transformers learning the one-nearest neighbor rule. A single softmax attention layer can successfully learn to behave like a one-nearest neighbor.
  arXiv  Detail & Related papers  (2024-11-16T16:12:42Z)
• Extracting Finite State Machines from Transformers [0.3069335774032178]
  We investigate the trainability of transformers trained on regular languages from a mechanistic interpretability perspective. We empirically find tighter lower bounds on the trainability of transformers, when a finite number of symbols determine the state. Our mechanistic insight allows us to characterise the regular languages a one-layer transformer can learn with good length generalisation.
  arXiv  Detail & Related papers  (2024-10-08T13:43:50Z)
• In-Context Learning with Representations: Contextual Generalization of Trained Transformers [66.78052387054593]
  In-context learning (ICL) refers to a capability of pretrained large language models, which can learn a new task given a few examples during inference. This paper investigates the training dynamics of transformers by gradient descent through the lens of non-linear regression tasks.
  arXiv  Detail & Related papers  (2024-08-19T16:47:46Z)
• What Can Transformer Learn with Varying Depth? Case Studies on Sequence Learning Tasks [15.874604623294427]
  We show a transformer with only one attention layer can excel in memorization but falls short in other tasks. We identify a class of simple operations that a single attention layer can execute, and show that the complex tasks can be approached as the combinations of these simple operations.
  arXiv  Detail & Related papers  (2024-04-02T02:45:12Z)
• How Transformers Learn Causal Structure with Gradient Descent [44.31729147722701]
  Self-attention allows transformers to encode causal structure. We introduce an in-context learning task that requires learning latent causal structure. We show that transformers trained on our in-context learning task are able to recover a wide variety of causal structures.
  arXiv  Detail & Related papers  (2024-02-22T17:47:03Z)
• Differentiable Subset Pruning of Transformer Heads [71.7904179689271]
  We introduce a new head pruning technique that we term differentiable subset pruning. We show that differentiable subset pruning performs comparably or better than previous works while offering precise control of the sparsity level.
  arXiv  Detail & Related papers  (2021-08-10T13:08:34Z)
• Thinking Like Transformers [64.96770952820691]
  We propose a computational model for the transformer-encoder in the form of a programming language. We show how RASP can be used to program solutions to tasks that could conceivably be learned by a Transformer. We provide RASP programs for histograms, sorting, and Dyck-languages.
  arXiv  Detail & Related papers  (2021-06-13T13:04:46Z)
• Multi-branch Attentive Transformer [152.07840447196384]
  We propose a simple yet effective variant of Transformer called multi-branch attentive Transformer. The attention layer is the average of multiple branches and each branch is an independent multi-head attention layer. Experiments on machine translation, code generation and natural language understanding demonstrate that such a simple variant of Transformer brings significant improvements.
  arXiv  Detail & Related papers  (2020-06-18T04:24:28Z)
• Fixed Encoder Self-Attention Patterns in Transformer-Based Machine Translation [73.11214377092121]
  We propose to replace all but one attention head of each encoder layer with simple fixed -- non-learnable -- attentive patterns. Our experiments with different data sizes and multiple language pairs show that fixing the attention heads on the encoder side of the Transformer at training time does not impact the translation quality.
  arXiv  Detail & Related papers  (2020-02-24T13:53:06Z)

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.