A new dataset named CCore GLB-3D-Coder has been released, containing 3,918 examples across 40 diverse domains for generating 3D assets. The dataset includes complete conversations, reasoning traces, and validated HTML/JavaScript responses, along with rendered reference images and static assets. It is licensed under MIT and comprises 3,501 training and 417 validation examples, intended for use in fine-tuning and developing 3D tools. AI
IMPACT Provides a new resource for training models focused on 3D asset generation, potentially accelerating development in this area.
RANK_REASON Release of a dataset for training a model, with associated code and weights. [lever_c_demoted from research: ic=1 ai=1.0]
AI-generated summary · Google Gemini · from 1 sources. How we write summaries →