jacob-valdez/synthux-economy-r5-w076
收藏资源简介:
SynthUX计算机使用数据集(经济模拟r5)是一个基于合成生成的视觉计算机使用轨迹数据集,专注于模拟公司环境中的工人设备会话。每个数据记录代表一个工人在模拟公司内的会话过程:目标扩展为节点树,终端节点通过低级别鼠标和键盘输入驱动真实的桌面模拟器应用程序(包括终端、笔记、VS Code、浏览器、Slack/Teams、邮件、表格、幻灯片等),观察到的轨迹被记录为屏幕视频和每个节点的帧。数据集支持应用原生多应用程序执行(无工作台覆盖),提供每个参与者内容的多样性、跨操作系统消息传递以及大型多公司经济模拟。每个经济模拟作为独立的数据集仓库发布,内容包括data.jsonl(每个渲染轨迹的记录)、screenshots/(每个终端节点的帧,以PNG格式存储)和videos/(完整屏幕录制,以WebM格式存储)。记录模式涵盖轨迹ID、参与者ID、公司、角色ID、标题、环境(模拟器类型如macos-web-next、windows-web-next或browser-os)、自然语言目标、节点树、密集低级别输入事件、因果模拟器状态观察、对齐信息以及媒体文件引用。本版本包含36个轨迹、1059帧、6个公司,按环境分类:browser-os(9个)、macos-web-next(11个)、windows-web-next(16个)。数据由SynthUX工具生成,观察结果基于因果模拟(突变通道为低级别输入),在捕获过程中未使用高级别应用程序突变。
The SynthUX Computer-Use Dataset — economy sim r5 is a synthetic dataset of grounded visual computer-use trajectories, focusing on worker device sessions within simulated company environments. Each record represents one workers session in a simulated company: a goal expands into a node tree, terminal nodes drive real desktop-simulator applications (such as Terminal, Notes, VS Code, Browser, Slack/Teams, Mail, Sheets, Slides, etc.) through low-level mouse and keyboard input, and the observed trajectory is recorded as a screen video plus per-node frames. The dataset supports app-native multi-application execution (without a workbench overlay), per-actor content diversity, cross-OS messaging, and large multi-company economy simulations. Each economy simulation is published as its own dataset repository. Contents include data.jsonl (one record per rendered trajectory), screenshots/ (frames per terminal node in PNG format), and videos/ (full screen recordings in WebM format). The record schema covers trajectory_id, actor_id, company, role_id, title, environment (simulator types: macos-web-next, windows-web-next, or browser-os), goal, tree.nodes, trajectory.input_events (dense low-level events with timestamps), trajectory.observations (causal simulator state), trajectory.alignment, media.video, and media.frames. This build contains 36 trajectories, 1059 frames, 6 companies, and is categorized by environment: browser-os (9), macos-web-next (11), windows-web-next (16). The data is generated with SynthUX, with observations being causal (mutation channel = low_level_input); no high-level app mutation is used during capture.




