cad-1000-hours

 # CAD 1000 Hours CAD 1000 Hours is a computer-use dataset containing 1,021.64 hours of recorded work across 597 workflows and 10 CAD, BIM, structural-analysis, and visualization applications. The tables below summarize its software coverage. ## Category distribution | Category | Included software | Workflows | Hours | Share of hours | |---|---|---:|---:|---:| | Drafting and general CAD | AutoCAD | 238 | 501.99 | 49.14% | | Mechanical and product CAD | SOLIDWORKS, CATIA, Siemens NX | 282 | 305.27 | 29.88% | | Architecture, BIM, and 3D modeling | SketchUp, Revit Architecture, Revit Structure | 67 | 203.82 | 19.95% | | Structural analysis | STAAD.Pro | 6 | 6.54 | 0.64% | | Visualization and rendering | V-Ray, D5 Render | 4 | 4.02 | 0.39% | | **Total** | **10 software packages** | **597** | **1,021.64** | **100.00%** | ## Software distribution | Software | Workflows | Hours | |---|---:|---:| | AutoCAD | 238 | 501.99 | | SOLIDWORKS | 149 | 209.04 | | SketchUp | 49 | 142.93 | | CATIA | 45 | 53.60 | | Revit Architecture | 15 | 51.78 | | Siemens NX | 88 | 42.63 | | Revit Structure | 3 | 9.11 | | STAAD.Pro | 6 | 6.54 | | V-Ray | 3 | 3.04 | | D5 Render | 1 | 0.98 | | **Total** | **597** | **1,021.64** | ## Dataset statistics | Metric | Value | |---|---:| | Recorded workflow time | 1,021.64 hours | | Workflows | 597 | | Software packages | 10 | | Dataset artifact files | 8,973 | | Dataset size | 256.60 GiB | ## Workflow artifacts Each workflow is self-contained and provides the recording, synchronized interaction data, task specification, source materials, and completed deliverables. | File or directory | Description | |---|---| | `clip.mp4` | Screen recording encoded at 30 FPS. | | `events.json` | Timestamped mouse and keyboard input events synchronized with the screen recording. | | `frame_events.json` | Frame-level timing data used to align video frames with recorded input events. | | `metadata.json` | Workflow-level recording and processing metadata. | | `narration.json` | Frame-level natural-language descriptions of the on-screen activity. | | `task_desc.json` | Task instructions and the expected deliverable. | | `rubrics.json` | Requirements and scoring criteria for evaluating the completed output. | | `task_overview.pdf` | Human-readable task overview with instructions and reference material. | | `input_files/` | Source and reference assets provided for completing the task. | | `output_files/` | Final CAD project files and exported deliverables produced by the workflow. | ## Repository organization Workflows are organized as `<software>/<workflow-id>/`. Files inside `input_files/` are stored directly in that directory, with no nested subdirectories. ```text <software>/<workflow-id>/ ├── clip.mp4 ├── events.json ├── frame_events.json ├── metadata.json ├── narration.json ├── task_desc.json ├── rubrics.json ├── task_overview.pdf ├── input_files/ └── output_files/ ```