Dataset · gnucleus-ai
AI agent run logs and results in FreeCAD (Bench v3)
gnucleus-ai/cad-gen-freecad-bench-v3
1,300 final attempts from 13 agent systems on a parametric 3D modeling benchmark in FreeCAD, complete with logs and results.
What's inside
Contains the results of 1,300 evaluated attempts (one final attempt per pair of 13 systems and tasks) for the Parametric CAD Bench v3 leaderboard. The dataset includes a summary table in parquet format, as well as detailed run folders containing configurations, execution logs, and final results. Tasks are published as runnable jobs for the Harbor environment.
What you can do with it
- Analyzing errors and trajectory patterns of AI agents in 3D model generation
- Comparing the performance of various LLMs and agent architectures in CAD design tasks
- Selecting successful trajectories for SFT fine-tuning of FreeCAD script generation models
- Size
- 1K–10K rows
- Downloads this month
- 92.8K
- Likes
- 1
How to load it
from datasets import load_dataset
ds = load_dataset("gnucleus-ai/cad-gen-freecad-bench-v3")