--- title: README emoji: 🦀 colorFrom: red colorTo: yellow sdk: static pinned: false --- # RustingStudio We build machine-learning software in Rust and train our own models on consumer hardware. Everything here was trained from scratch on a single RTX 3060 12 GB, with no Python in training or in use. The code is open: - [RustingBrain](https://github.com/RustingStudio/RustingBrain): the deep-learning library the models are trained with. - [RustingRunner](https://github.com/RustingStudio/RustingRunner): the small program that runs the released weights. - [RustingStrata](https://github.com/RustingStudio/RustingStrata): an inference engine for Qwen GGUF models. ## Models **[Kaze-3D-v0.1](https://huggingface.co/rusting-studio/Kaze-3D-v0.1)**: one image in, one 3D mesh out, in 4 seconds on an RTX 3060. An early release: it gets the rough shape of an object, geometry only. Apache-2.0, trained only on CC-BY and CC0 data. The model page has the measured numbers and the limits. More at [rustingstudio.com](https://rustingstudio.com) and [github.com/RustingStudio](https://github.com/RustingStudio).