Submissions and evaluation results for the CADGenBench leaderboard.
Leaderboard Space: HuggingAI4Engineering/CADGenBench.
Benchmark code: github.com/huggingface/cadgenbench.
Fixture inputs: HuggingAI4Engineering/cadgenbench-data.
Ground truth: HuggingAI4Engineering/cadgenbench-data-gt (private).
results.jsonl
One row per submitted + evaluated submission. The leaderboard table reads from this.… See the full description on the dataset page:
https://huggingface.co/datasets/HuggingAI4Engineering/cadgenbench-submissions.