Hacker News new | ask | show | jobs
by ac29 33 days ago
The inference providers are running batch sizes much larger than 10