Kieran O'Connor

A massive, highly-diverse dataset of Fused Deposition Modeling (FDM) G-codes generated directly from Printables. This dataset is specifically designed for training machine learning models on raw 3D printing manufacturing instructions (G-code). It can be used for tasks like G-code generation, print failure prediction, semantic analysis of toolpaths, and printer-agnostic slice classification. To ensure a highly robust and diverse set of training data, every original 3D model from the source dataset has been iteratively sliced into multiple variants (default: 10 per model) using PrusaSlicer. The pipeline supports and natively slices.stl,.obj,.3mf,.step, and.amf files. For each variant, a…