Run GPU-accelerated Docker containers for cost-effective inference   Link: GPUX.AI