Run a language model on your AMD Instinct GPU.
Recent revisions are kept. Restoring one creates a new revision.
