Re:OpenAI Is Now Everything It Promised Not to Be: Closed-So...
by bitL on Wednesday March 01, 2023 @09:14AM
(#34982598)
Huh, how would an enthusiast get 1000x H100 GPUs to train some state-of-art model? ML workflows are the most expensive computing pipelines these days. It's easy to spend $1M/month just on inference alone in production. Without financial backing there would be some trivial useless models everywhere (there already are).
[ Parent
| on HN ]
There are no comments on this comment yet.