llm-inference Contributions to accelerate and scale LLM inferences. For now, a simulator of vLLM scheduling strategy.