Run, tune, and scale generative models with OctoAI’s compute service