tldr

该数据集是一个包含自然语言意图与Bash命令行配对的标准基准，旨在帮助用户从自然语言描述中生成Bash命令。每个自然语言意图都对应一个单一的Bash命令，该基准的建立旨在提高代码生成模型在生成Bash命令方面的泛化能力。数据集的规模包括1,879个独特的Bash命令和9,187个自然语言到Bash命令的配对。这项任务的目的是为Bash命令的自然语言到代码生成提供支持。

This dataset is a standard benchmark comprising pairs of natural language intents and Bash command lines, designed to help users generate Bash commands from natural language descriptions. Each natural language intent corresponds to exactly one single Bash command. This benchmark was developed to enhance the generalization capability of code generation models when producing Bash commands. The dataset consists of 1,879 unique Bash commands and 9,187 natural language-to-Bash command pairs. The purpose of this task is to support natural language-to-code generation for Bash commands.