Y
Hacker News
new
|
ask
|
show
|
jobs
by
nick0garvey
1013 days ago
This isn't unreasonable. ML workloads benefit from more computational time per request. Lower QPS = better results.