Description
Local language models can run through a development build of Ollama with ROCm support. This helps users experiment with AMD GPU acceleration and newer upstream changes.
It can download models, expose a local API, and use significant GPU resources. Development builds may change quickly, so verify compatibility before important workflows.