george_123 | Dark Hacker News

1.	Loading Llama-2 70B 20x faster with Anyscale Endpoints(anyscale.com) 3 points by george_123 2 years ago \| 0 comments
2.	How continuous batching improves LLM inference throughput 23x(twitter.com) 1 points by george_123 3 years ago \| 0 comments
3.	Ant Group – scaling to 1.37M QPS on Ray(anyscale.com) 3 points by george_123 3 years ago \| 1 comment
4.	1 points by george_123 3 years ago \| discuss