B blog.vllm.ai vLLM v0.6.0: 2.7x Throughput Improvement and 5x Latency Reduction 3 points xmo 2 years ago 0 comments