CADFS: A Large-Scale Dataset for Parametric CAD Model Generation

CADFS is a large-scale dataset for parametric CAD model generation from text descriptions and multi-view images. Models are represented as FeatureScript programs, enabling direct import into the Onshape environment. The dataset was created by VladPyatov and was used to train and evaluate the CADFS-2B multimodal language model.

Use Cases

• Train multimodal language models for CAD program generation based on text descriptions.
• Evaluate model performance on parametric CAD generation tasks based on multi-view images.
• Generate FeatureScript programs for direct use in the Onshape CAD environment.

Strengths

• Designed as a large-scale dataset for a complex multimodal task.
• Models are represented as FeatureScript programs, enabling direct use in a professional CAD environment.

Limitations

• Description metadata is limited; actual data quality requires manual inspection after download.
• Column-level documentation is absent; field semantics must be inferred after download.
• Row count is unknown, which may limit suitability assessment.

Provenance

Source

huggingface

Freshness

Last updated 2026-06-01 10:11:52; freshness should be verified.