Y
HN Search
Hacker News Search
new
|
comments
|
top
|
jobs
loading thread…
3 ms
·
Continuous batching enables 23x throughput in LLM inference
2 points
by
richardliaw
3y ago