TuneOps Desktop (Alpha)
TuneOps Desktop is a standalone macOS application for fine-tuning open foundation models on your own hardware. It’s built for machine learning practitioners and enthusiasts who want to try LoRA adapters and small fine-tunes without provisioning cloud GPUs or hand-rolling a Python, CUDA, and Kubernetes stack.
How it works
Point the app at a base model on Hugging Face, point it at a dataset, pick a training recipe, and click Run. The app downloads the model and dataset, schedules the training job, streams logs and metrics into a native UI, and when the run finishes you can chat with the resulting adapter in-app.
What’s under the hood
Tuning runs on your own hardware. Each run is scheduled as a job on a local Kubernetes cluster that the app manages for you, and the tuning itself runs in PyTorch. Nothing about your model, dataset, or run history leaves the machine unless you explicitly send it somewhere.
Alpha scope
The first release is macOS. Linux and Windows are on the roadmap for later releases.
Preview access
The Desktop app is in private Alpha. If you want to try it on your own models and datasets, tell us what you’re working on and we’ll send you a build.