让模型在生成序列时可选择性地让专家接手部分输出，提升效率与准确率。
Learning to Partially Defer for Sequences
- 针对长序列预测，提出逐标记细化拒答机制，而非整体拒绝
- 实验显示在旅行商、新闻摘要等任务上，粒度化拒答更优
- 适用于需专家介入的复杂序列生成场景
在学习拒答（L2D）框架中，预测模型可选择自主预测或交由专家处理，由拒答器决定。现有方法训练拒答器对整个预测结果进行拒绝，这在处理长序列时并不理想。本文提出一种面向序列输出的L2D新设定，允许系统将模型预测中的特定输出片段交给专家，实现机器与专家的交替协作。我们设计两类基于模型的后处理拒答器：一种是逐标记拒答器，可将单个标记的预测交给具备下一标记预测能力的专家；另一种是一次性拒答器，适用于无此能力的专家，用于从某一点起全部交由专家完成剩余序列。实验表明，在旅行商求解、新闻摘要和天气预测任务中，这种细粒度拒答策略相比整体拒答实现了更优的成本-准确率权衡。
原文摘要 · Abstract (English)
In the Learning to Defer (L2D) framework, a prediction model can either make a prediction or defer it to an expert, as determined by a rejector. Current L2D methods train the rejector to decide whether to reject the {\em entire prediction}, which is not desirable when the model predicts long sequences. We present an L2D setting for sequence outputs where the system can defer \textit{specific outputs} of the whole model prediction to an expert in an effort to interleave the expert and machine throughout the prediction. We propose two types of model-based post-hoc rejectors for pre-trained predictors: a token-level rejector, which defers specific token predictions to experts with next token prediction capabilities, and a one-time rejector for experts without such abilities, which defers the remaining sequence from a specific point onward. In the experiments, we also empirically demonstrate that such granular deferrals achieve better cost-accuracy tradeoffs than whole deferrals on Traveling salesman solvers, News summarization, and Weather prediction.
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。