Redirecting to Faster LLM Inference via Sequential Monte Carlo...