A model of errors in transformers

• URL: http://arxiv.org/abs/2601.14175v1
• Date: Tue, 20 Jan 2026 17:27:03 GMT
• Title: A model of errors in transformers
• Abstract summary: We study the error rate of LLMs on tasks that require a deterministic output, and repetitive processing of tokens drawn from a small set of alternatives.<n>We argue that incorrect predictions arise when small errors in the attention mechanism accumulate to cross a threshold.<n>We show how to construct prompts to reduce the error rate.
• Score: 14.482123927397135
• License: http://creativecommons.org/licenses/by/4.0/
• Abstract: We study the error rate of LLMs on tasks like arithmetic that require a deterministic output, and repetitive processing of tokens drawn from a small set of alternatives. We argue that incorrect predictions arise when small errors in the attention mechanism accumulate to cross a threshold, and use this insight to derive a quantitative two-parameter relationship between the accuracy and the complexity of the task. The two parameters vary with the prompt and the model; they can be interpreted in terms of an elementary noise rate, and the number of plausible erroneous tokens that can be predicted. Our analysis is inspired by an ``effective field theory'' perspective: the LLM's many raw parameters can be reorganized into just two parameters that govern the error rate. We perform extensive empirical tests, using Gemini 2.5 Flash, Gemini 2.5 Pro and DeepSeek R1, and find excellent agreement between the predicted and observed accuracy for a variety of tasks, although we also identify deviations in some cases. Our model provides an alternative to suggestions that errors made by LLMs on long repetitive tasks indicate the ``collapse of reasoning'', or an inability to express ``compositional'' functions. Finally, we show how to construct prompts to reduce the error rate.

Related papers

• Deep Interaction: An Efficient Human-AI Interaction Method for Large Reasoning Models [16.90017128904664]
  We propose an efficient human intervention mechanism for precisely correcting reasoning errors in large language models (LLMs)<n>Our approach enables direct editing of the original response, allowing erroneous parts to be corrected while preserving accurate reasoning steps.<n>Our method achieves over a 25% improvement in correction success rate and reduces token usage by approximately 40% on STEM tasks reasoning.
  arXiv  Detail & Related papers  (2026-07-15T17:16:43Z)
• Emergence of Minimal Circuits for Indirect Object Identification in Attention-Only Transformers [0.10152838128195467]
  We train small, attention-only transformers from scratch on a symbolic version of the Indirect Object Identification task.<n>A single-layer model with only two attention heads achieves perfect IOI accuracy, despite lacking residuals and normalization layers.<n>A two-layer, one-head model achieves similar performance by composing information across layers through query-value interactions.
  arXiv  Detail & Related papers  (2025-10-28T22:25:19Z)
• Unveiling Reasoning Thresholds in Language Models: Scaling, Fine-Tuning, and Interpretability through Attention Maps [3.8936716676293917]
  This study investigates the in-context learning capabilities of various decoder-only transformer-based language models with different model sizes and training data.<n>We identify a critical parameter threshold (1.6 billion), beyond which reasoning performance improves significantly in tasks such as commonsense reasoning in multiple-choice question answering and deductive reasoning.
  arXiv  Detail & Related papers  (2025-02-21T00:48:32Z)
• Error Classification of Large Language Models on Math Word Problems: A Dynamically Adaptive Framework [79.40678802098026]
  Math Word Problems serve as a crucial benchmark for evaluating Large Language Models' reasoning abilities.<n>Current error classification methods rely on static and predefined categories.<n>We propose Error-Aware Prompting (EAP) that incorporates common error patterns as explicit guidance.
  arXiv  Detail & Related papers  (2025-01-26T16:17:57Z)
• Subtle Errors in Reasoning: Preference Learning via Error-injected Self-editing [59.405145971637204]
  We propose a novel preference learning framework called eRror-Injected Self-Editing (RISE)<n>RISE injects predefined subtle errors into pivotal tokens in reasoning or steps to construct hard pairs for error mitigation.<n>Experiments validate the effectiveness of RISE, with preference learning on Qwen2-7B-Instruct yielding notable improvements of 3.0% on GSM8K and 7.9% on MATH with only 4.5K training samples.
  arXiv  Detail & Related papers  (2024-10-09T07:43:38Z)
• A Coin Has Two Sides: A Novel Detector-Corrector Framework for Chinese Spelling Correction [79.52464132360618]
  Chinese Spelling Correction (CSC) stands as a foundational Natural Language Processing (NLP) task. We introduce a novel approach based on error detector-corrector framework. Our detector is designed to yield two error detection results, each characterized by high precision and recall.
  arXiv  Detail & Related papers  (2024-09-06T09:26:45Z)
• Unveiling the Statistical Foundations of Chain-of-Thought Prompting Methods [59.779795063072655]
  Chain-of-Thought (CoT) prompting and its variants have gained popularity as effective methods for solving multi-step reasoning problems. We analyze CoT prompting from a statistical estimation perspective, providing a comprehensive characterization of its sample complexity.
  arXiv  Detail & Related papers  (2024-08-25T04:07:18Z)
• Understanding and Mitigating Classification Errors Through Interpretable Token Patterns [58.91023283103762]
  Characterizing errors in easily interpretable terms gives insight into whether a classifier is prone to making systematic errors. We propose to discover those patterns of tokens that distinguish correct and erroneous predictions. We show that our method, Premise, performs well in practice.
  arXiv  Detail & Related papers  (2023-11-18T00:24:26Z)
• Evaluating Robustness to Dataset Shift via Parametric Robustness Sets [7.347989843033034]
  We give a method for proactively identifying small, plausible shifts in distribution which lead to large differences in model performance. We apply an approach to classifying gender from images, revealing sensitivity to shifts in non-causal attributes.
  arXiv  Detail & Related papers  (2022-05-31T16:44:18Z)
• Localized Debiased Machine Learning: Efficient Inference on Quantile Treatment Effects and Beyond [69.83813153444115]
  We consider an efficient estimating equation for the (local) quantile treatment effect ((L)QTE) in causal inference. Debiased machine learning (DML) is a data-splitting approach to estimating high-dimensional nuisances. We propose localized debiased machine learning (LDML), which avoids this burdensome step.
  arXiv  Detail & Related papers  (2019-12-30T14:42:52Z)

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.