上下文内分类任务中数据依赖与突跳式学习的机制基础
The mechanistic basis of data dependence and abrupt learning in an in-context classification task
AI总结:
本文用最小仅注意力网络复现数据分布对 ICL 与权重内学习的调控,提出双参数归纳头模型,解释其经嵌套 logit 顺序学习而突现的机制。
摘要:
Transformer 模型表现出上下文内学习(in-context learning)能力:能够依据输入序列中的示例,准确预测对新查询的响应。上下文内学习与传统的、把查询—输出关系学习到权重中的 in-weights learning 形成对比。训练数据分布和架构的哪些方面会促进上下文内学习而非权重内学习?近期研究表明,语言固有的特定分布属性,例如突发性、大型词典以及偏斜的秩—频率分布,会控制这两种学习形式之间的权衡或同时出现。我们首先表明,在一个经过简化数据集训练的最小化仅注意力网络中,这些结果会复现。上下文内学习(ICL)由归纳头(induction head)的突然出现所驱动,该归纳头随后与权重内学习竞争。通过识别先于上下文内学习出现的进展度量,并开展有针对性的实验,我们构建了一个归纳头的双参数模型,该模型模拟了基于注意力的网络所展现的完整数据分布依赖关系。一个关于归纳头形成的现象学模型,将其突然出现追溯到由内在课程所促成的三个嵌套 logit 的顺序学习。我们提出,基于注意力的网络中的尖锐转变,源于实现 ICL 所必需的一条特定多层操作链;该链由训练期间顺序学习到的嵌套非线性所实现。
英文摘要:
Transformer models exhibit in-context learning: the ability to accurately predict the response to a novel query based on illustrative examples in the input sequence. In-context learning contrasts with traditional in-weights learning of query-output relationships. What aspects of the training data distribution and architecture favor in-context vs in-weights learning? Recent work has shown that specific distributional properties inherent in language, such as burstiness, large dictionaries and skewed rank-frequency distributions, control the trade-off or simultaneous appearance of these two forms of learning. We first show that these results are recapitulated in a minimal attention-only network trained on a simplified dataset. In-context learning (ICL) is driven by the abrupt emergence of an induction head, which subsequently competes with in-weights learning. By identifying progress measures that precede in-context learning and targeted experiments, we construct a two-parameter model of an induction head which emulates the full data distributional dependencies displayed by the attention-based network. A phenomenological model of induction head formation traces its abrupt emergence to the sequential learning of three nested logits enabled by an intrinsic curriculum. We propose that the sharp transitions in attention-based networks arise due to a specific chain of multi-layer operations necessary to achieve ICL, which is implemented by nested nonlinearities sequentially learned during training.
机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。