marin-community/synth-machine-records-ppl

该数据集是一个多配置数据集，专注于各种系统和应用日志与配置文件的文本数据，用于自然语言处理任务。它包含15个独立配置，覆盖以下类别：CI日志（如GitHub Actions作业）、配置清单（如Dockerfile、GitHub Actions YAML、Kubernetes YAML、package.json、pyproject.toml、Terraform HCL）、服务日志（如Kubernetes系统事件、Nginx访问错误、systemd日志、Zeek连接HTTP/DNS日志）、结构化日志（JSON应用日志）和错误跟踪（如编译器错误、Python回溯）。每个配置的验证集包含1000个示例，特征包括输入文本、目标文本、ID、子集、任务、种子和元数据（如家族、配置名称等）。数据集可能用于文本生成、分类或分析任务，但具体用途未在README中说明。

This dataset is a multi-configuration dataset focusing on text data from various system and application logs and configuration files for natural language processing tasks. It includes 15 independent configurations covering the following categories: CI logs (e.g., GitHub Actions jobs), configuration manifests (e.g., Dockerfile, GitHub Actions YAML, Kubernetes YAML, package.json, pyproject.toml, Terraform HCL), service logs (e.g., Kubernetes system events, Nginx access errors, systemd journals, Zeek connection HTTP/DNS logs), structured logs (JSON application logs), and error traces (e.g., compiler errors, Python tracebacks). Each configurations validation split contains 1000 examples, with features including input text, target text, ID, subset, task, seed, and metadata (e.g., family, configuration name). The dataset is likely intended for text generation, classification, or analysis tasks, though specific usage is not detailed in the README.