Unsloth Studio
A graphical fine-tuning tool provided by Unsloth that simplifies the model fine-tuning workflow and manages model cards and exports.
Timeline (last 90 days)
The training, testing, export, and deployment workflow for decision models has been integrated into Unsloth Studio.
AMD ROCm Docker images with Unsloth Studio and JupyterLab have been added.
The new Studio's data is stored on a separate volume, so user data will not be lost when the container is updated.
The new version introduces a preview OS-level tool sandbox on supported Linux and macOS hosts, but network access remains unrestricted
Unsloth Studio has introduced a multi-user account system in which each account's chat history, projects, credentials, and training data are isolated from one another, while allowing already-loaded models to be shared when configurations match.
Unsloth Studio adds support for the standard input_audio message format and fixes an issue where API keys longer than 190 bytes were incorrectly rejected.
Unsloth Desktop and the web-based Studio are completely consistent in core functionality; the desktop version is a packaged version designed to simplify use for general users.
All Facts (7)
The new version introduces a preview OS-level tool sandbox on supported Linux and macOS hosts, but network access remains unrestricted
70%UnverifiedThe training, testing, export, and deployment workflow for decision models has been integrated into Unsloth Studio.
50%UnverifiedThe new Studio's data is stored on a separate volume, so user data will not be lost when the container is updated.
50%UnverifiedUnsloth Studio has introduced a multi-user account system in which each account's chat history, projects, credentials, and training data are isolated from one another, while allowing already-loaded models to be shared when configurations match.
50%UnverifiedUnsloth Studio adds support for the standard input_audio message format and fixes an issue where API keys longer than 190 bytes were incorrectly rejected.
50%UnverifiedUnsloth Desktop and the web-based Studio are completely consistent in core functionality; the desktop version is a packaged version designed to simplify use for general users.
50%UnverifiedAMD ROCm Docker images with Unsloth Studio and JupyterLab have been added.
70%