Aha.
正在载入中英对照阅读…

arXiv:2303.10475 · 中英对照阅读

Large Language Model Instruction Following: A Survey of Progresses and Challenges

Renze Lou、Kai Zhang、Wenpeng Yin

中文速览

当任务缺少标注样本、又希望系统能快速上手新任务时,怎样让模型靠文字说明而不是大量例子学会做事?这篇综述把相关研究放在“间接监督”的框架下梳理,比较了面向分类推断、面向语言模型的提示,以及面向人的详细指令,并总结模型如何学习指令、常用数据集和评测方法。综述发现,模型更大、训练任务更多样,通常更能举一反三;但现有提示往往不够好懂,也难覆盖复杂任务,详细指令则更易理解、应用面广,却更难让模型学会。它的重要性在于把分散的指令学习研究串成一幅全景图,帮助研究者看清哪些方法适合什么任务,以及接下来该补上哪些短板。

摘要

Task semantics can be expressed by a set of input-output examples or a piece of textual instruction. Conventional machine learning approaches for natural language processing (NLP) mainly rely on the availability of large-scale sets of task-specific examples. Two issues arise: first, collecting task-specific labeled examples does not apply to scenarios where tasks may be too complicated or costly to annotate, or the system is required to handle a new task immediately; second, this is not user-friendly since end-users are probably more willing to provide task description rather than a set of examples before using the system. Therefore, the community is paying increasing interest in a new supervision-seeking paradigm for NLP: learning to follow task instructions, i.e., instruction following. Despite its impressive progress, there are some unsolved research equations that the community struggles with. This survey paper tries to summarize and provide insights to the current research on instruction following, particularly, by answering the following questions: (i) What is task instruction, and what instruction types exist? (ii) How to model instructions? (iii) What are popular instruction following datasets and evaluation metrics? (iv) What factors influence and explain the instructions’ performance? (v) What challenges remain in instruction following? To our knowledge, this is the first comprehensive survey about instruction following.11 1 The curated paper list can be found at: https://github.com/RenzeLou/awesome-instruction-learning

术语表

task semantics
任务语义
task-specific examples
特定任务示例
instruction following
指令遵循
task instruction
任务指令
instruction types
指令类型
instruction-driven NLP
指令驱动的自然语言处理
large language models (LLMs)
大语言模型(LLMs)
prompt
提示
prompt learning
提示学习
in-context instruction
上下文内指令
input-by-output demonstrations
输入—输出示例
chain-of-thought reasoning
思维链推理
supervision-seeking paradigm
寻求监督范式
indirect supervision
间接监督
example-driven supervised learning
示例驱动的监督学习
instruction tuning
指令微调
instruction-tuning dataset
指令微调数据集
few-shot learning
少样本学习
zero-shot learning
零样本学习
downstream task
下游任务
task generalization
任务泛化
semantic parsing
语义解析
parameter-efficient fine-tuning
参数高效微调
soft instruction
软指令
evaluation metric
评测指标
model scale
模型规模
dual-track scaling
双轨扩展
input (X)
输入(X)
output (Y)
输出(Y)
template (T)
模板(T)