Continuous batching enables 23x throughput in LLM inference (anyscale.com)

2 points by richardliaw 3y ago ↗ HN

0 comments

[ 2.8 ms ] story [ 11.6 ms ] thread

No comments yet.

A showcase of shard-db. Data from the Hacker News API. Refreshed every 15 minutes.

Source · Live DB stats