For AI agents: The documentation index is at https://docs.digitalocean.com/llms.txt. Markdown versions of pages use the same URL with index.html.md in place of the HTML page (for example, append index.html.md to the directory path instead of opening the HTML document).

Unsloth Studio on AMD ROCm 7.2.4

This 1-Click image ships Unsloth Studio 2026.7.2, a web based environment for fast, memory efficient fine-tuning of large language models, running on an AMD ROCm 7.2.4 host stack on Ubuntu 24.04. It is preconfigured for AMD Instinct GPUs.

Unsloth accelerates LoRA and QLoRA fine-tuning of popular open models (Llama, Qwen, Mistral, Gemma and more) with significantly reduced VRAM usage, and supports 4-bit quantized training out of the box. Unsloth Studio provides a browser based UI to load models, prepare datasets, configure and launch training runs, and compare results, without writing boilerplate.

The image bundles PyTorch 2.11.0 built for ROCm 7.2, so training runs use native AMD GPU acceleration end to end.

Software Included

Deploying this Offering using the Control Panel

Click the Deploy to DigitalOcean button to deploy this offering. If you aren’t logged in, this link will prompt you to log in with your DigitalOcean account.

[](https://cloud.digitalocean.com/gpus/new?appId=acc39cfbc45ef9726190f532&image=Unsloth Studio (ROCm 7.2.4) 2026.7.2 on Ubuntu 24.04&type=applications)

Getting Started After Deploying Unsloth Studio (ROCm 7.2.4)

After the droplet boots, open http:// in your browser. Unsloth Studio serves on port 80 and prompts you to create an admin password on first access. Set this password immediately to prevent unauthorized access. You can also SSH in as root; the login message shows how to inspect the unsloth-studio service and logs. Verify the GPU stack with amd-smi and check the pinned Unsloth version inside the Studio virtualenv at /root/.unsloth/studio.