jacob-valdez/synthux-economy-r5-w032
收藏资源简介:
SynthUX计算机使用数据集(经济模拟r5)是一个由SynthUX生成的基于视觉的计算机使用轨迹数据集。每个记录代表一个工人在模拟公司内的设备会话:目标扩展为节点树,终端节点通过低级别的鼠标/键盘输入驱动真实的桌面模拟器应用程序(包括Terminal、Notes、VS Code、Browser、Slack/Teams、Mail、Sheets、Slides等),观察到的轨迹被记录为屏幕视频加上每个节点的帧。该数据集支持应用原生多应用程序执行(无工作台覆盖)、每个参与者的内容多样性、跨操作系统消息传递和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。数据集包含data.jsonl文件(每个渲染轨迹一个记录)、screenshots文件夹(每个轨迹的终端节点帧,以PNG格式存储)和videos文件夹(完整的屏幕录制,以WebM格式存储)。记录模式包括trajectory_id、actor_id、company、role_id、title、env、goal、tree.nodes、trajectory.input_events、trajectory.observations、trajectory.alignment、media.video和media.frames等字段。统计信息显示有36个轨迹、896个帧、6个公司,并按环境(browser-os: 10, macos-web-next: 13, windows-web-next: 13)分布。数据集由SynthUX生成,观察是因果的(突变通道为低级别输入),在捕获过程中未使用高级应用程序突变。
The SynthUX Computer-Use Dataset (economy sim r5) is a dataset of grounded visual computer-use trajectories generated by SynthUX. Each record represents one workers device session inside 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/keyboard input, and the observed trajectory is recorded as a screen video plus per-node frames. It features app-native multi-application execution (no workbench overlay), per-actor content diversity, cross-OS messaging, and large multi-company economies. Each economy simulation is published as its own dataset repository. The dataset includes data.jsonl (one record per rendered trajectory), screenshots folder (frames per terminal node in PNG format), and videos folder (full screen recordings in WebM format). The record schema includes fields like trajectory_id, actor_id, company, role_id, title, env, goal, tree.nodes, trajectory.input_events, trajectory.observations, trajectory.alignment, media.video, and media.frames. Statistics show 36 trajectories, 896 frames, 6 companies, and distribution by environment (browser-os: 10, macos-web-next: 13, windows-web-next: 13). The dataset is generated with SynthUX, with observations being causal (mutation channel = low_level_input) and no high-level app mutation used during capture.




