kaito should support cpu based (amd64 and arm64) inference

vllm related docs:
https://docs.vllm.ai/en/v0.8.4/getting_started/installation/cpu.html
https://docs.vllm.ai/en/latest/models/supported_models.html