[Xaena]

Interesting benchmarks. I'm surprised vLLM is performing so poorly given all the corporate investment/attention it's getting. Is it because the contributions are more distributed than the average LLM team?