Selective Induction Heads: How Transformers Select Causal Structures In Context
选择性归纳头：Transformer 如何在上下文中选择因果结构AI 解读 ↓
← 返回论文列表与筛选
阅读说明
本页整理公开书目信息，不生成学术质量评分。AI 解读根据题名及所注明的摘要或原文片段生成，未完成全文核验，可能有误；具体假设、证明与结论请核对原文。摘要保留来源文本，未经模型改写。
DOI 首次登记不等于接收或发表日期。部分更正、评论等条目没有独立摘要；缺少可靠材料时保留待补状态。
本页为预印本版本；arXiv 收录、作者填写的期刊信息以及作者既有发表经历，都不单独证明本篇已接收。作者发表记录按 OpenAlex 身份与完整姓名、本站覆盖期刊 ISSN 及正式卷期核对，只展示已核实例，档案仍可能误合并。
引用量采用 OpenAlex 单条关联记录，不合计多个记录；关联记录可能包含正式发表版本，不等于此版本单独引用。缺失不视为零，引用指标不代表论文质量。
关注作者按书目中的完整姓名匹配，同名和姓名变体需人工辨别。相关论文先按共享细分方向和具体方法排列，再参考通用方法与发表日期。
分类依据与纠错
AI 根据所提供的论文材料归类：摘要研究 Transformer 的序列学习电路，未提供经济因果推断或所列领域应用的直接依据。
反馈分类问题
正在核对关联版本…
代码、数据与补充材料
正在核对对应版本的资源…
选择性归纳头：Transformer 如何在上下文中选择因果结构
- 研究问题
- 当序列依赖的滞后结构随上下文改变、而转移概率保持固定时，Transformer 能否识别正确结构并据此预测下一个词元？
- 方法与思路
- 作者以不同滞后的交错马尔可夫链构造任务，研究选择性归纳头，并给出实现该机制的三层 Transformer 构造及理论分析。
- 材料陈述的贡献
- 摘要称实验观察到模型选择正确滞后并复制历史词元，理论分析表明所构造机制渐近收敛到最大似然解，帮助解释模型行为。
- 阅读关注点
- 应核对交错序列生成规则、注意力电路的具体构造和渐近结论条件，并判断合成任务中的结构选择能否迁移到自然语言。
- 阅读边界
- 所给版本为 arXiv 预印本，同行评审及期刊接收未核验；摘要中的因果结构指序列依赖，未建立干预效应识别结论。
原文摘要
Transformers have exhibited exceptional capabilities in sequence modeling tasks, leveraging self-attention and in-context learning. Critical to this success are induction heads, attention circuits that enable copying tokens based on their previous occurrences. In this work, we introduce a novel framework that showcases transformers' ability to dynamically handle causal structures. Existing works rely on Markov Chains to study the formation of induction heads, revealing how transformers capture causal dependencies and learn transition probabilities in-context. However, they rely on a fixed causal structure that fails to capture the complexity of natural languages, where the relationship between tokens dynamically changes with context. To this end, our framework varies the causal structure through interleaved Markov chains with different lags while keeping the transition probabilities fixed. This setting unveils the formation of Selective Induction Heads, a new circuit that endows transformers with the ability to select the correct causal structure in-context. We empirically demonstrate that transformers learn this mechanism to predict the next token by identifying the correct lag and copying the corresponding token from the past. We provide a detailed construction of a 3-layer transformer to implement the selective induction head, and a theoretical analysis proving that this mechanism asymptotically converges to the maximum likelihood solution. Our findings advance the understanding of how transformers select causal structures, providing new insights into their functioning and interpretability.