Gear Design Dataset

Dataset Overview

The Gear Design Dataset contains structured data that includes gear design specifications and their corresponding DXF (Drawing Exchange Format) files. The dataset is intended for training and evaluating machine learning models focused on generative design and CAD (Computer-Aided Design) tasks. The dataset is split into three parts: training, validation, and test sets.

Each entry in the dataset represents a specific gear type and includes:

A prompt describing the gear specifications.
A corresponding DXF content (representing the 2D drawing of the gear).

The dataset is designed to support the development of generative models that can produce CAD designs from textual descriptions.

Data Splits

Train: The training set contains a variety of gear types with their corresponding prompts and DXF content.
Validation: The validation set is used for model evaluation during training.
Test: The test set is used to assess model performance after training.

Dataset Structure

The dataset contains the following fields:

gear_type: The type of gear (e.g., spur gear, bevel gear).
split: The dataset split, which can be "train", "val", or "test".
prompt: A textual description of the gear design specifications.
dxf_content: The corresponding DXF file content in string format.

Example:

{
  "gear_type": "spur gear",
  "split": "train",
  "prompt": "Generate a DXF file for a gear which conforms to the following description: Pressure Angle: 20\u00b0, Bending Strength (Nm): 111.21, Number of Teeth: 36, Pitch Diameter (mm): 90.0, Bore Diameter (mm): 32 with keyway, Surface Durability (Nm): 14.18, Material: SUS303, Type: Spur Gears, Tooth Finish: Cut.",
  "dxf_content": "<DXF file content>"
}