jacob-valdez/synthux-economy-r5-w062
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SynthUX生成的视觉计算机使用轨迹数据集,用于模拟公司环境中的工人设备会话。每个记录代表一个会话,其中目标扩展为节点树,终端节点通过低级鼠标/键盘输入驱动真实桌面模拟器应用程序(如Terminal、Notes、VS Code、Browser、Slack/Teams、Mail、Sheets、Slides等),观察轨迹以屏幕视频和每节点帧的形式记录。数据集支持应用程序原生多应用程序执行(无工作台覆盖)、每演员内容多样性、跨操作系统消息传递和大规模多公司经济模拟。每个经济模拟作为独立的数据集仓库发布。内容包含data.jsonl(每个渲染轨迹的记录)、screenshots文件夹(每个终端节点的帧图像)和videos文件夹(完整屏幕录制视频)。记录模式包括轨迹ID、演员ID、公司、角色、标题、环境(如macos-web-next、windows-web-next、browser-os)、自然语言目标、节点树、输入事件、观察状态、对齐信息以及媒体引用。本构建包含36个轨迹、834帧、6个公司,环境分布为browser-os(10个)、macos-web-next(8个)、windows-web-next(18个)。数据生成使用SynthUX(synthux.multicompany.day.v1),观察基于因果模拟(突变通道为低级输入),捕获过程中未使用高级应用程序突变。
The SynthUX Computer Usage Dataset (Economic Simulation r5) is a visual computer usage trajectory dataset generated via SynthUX, designed to simulate worker device sessions within corporate environments. Each record represents a session, where the target is expanded into a node tree, and terminal nodes drive real desktop applications (including Terminal, Notes, VS Code, Browser, Slack/Teams, Mail, Sheets, Slides, etc.) via low-level mouse and keyboard inputs. Interaction trajectories are recorded in the form of screen videos and per-node frame images. This dataset supports native multi-application execution without workspace overlay, per-actor content diversity, cross-operating system message passing, and large-scale multi-company economic simulations. Each economic simulation is released as an independent dataset repository. The dataset contents include data.jsonl (records for each rendered trajectory), the screenshots folder (frame images for each terminal node), and the videos folder (full-screen recorded videos). The record schema includes trajectory ID, actor ID, company, role, title, environment (e.g., macos-web-next, windows-web-next, browser-os), natural language target, node tree, input events, observation status, alignment information, and media references. This specific build contains 36 trajectories, 834 frames, and 6 companies, with an environment distribution of browser-os (10), macos-web-next (8), and windows-web-next (18). The data was generated using SynthUX (synthux.multicompany.day.v1), with observations based on causal simulation (with the mutation channel being low-level inputs), and no high-level application mutations were utilized during the capture process.




