52 points | by robertnishihara 11 hours ago
2 comments
Impressive performance work. It's interesting that you still see these 40+% perf gains like this.
Makes you think that you will continue to see the costs for a fixed level of "intelligence" dropping.
Love vLLM!
Impressive performance work. It's interesting that you still see these 40+% perf gains like this.
Makes you think that you will continue to see the costs for a fixed level of "intelligence" dropping.
Love vLLM!