logoalt Hacker News

antupis • today at 6:42 PM • 0 replies • view on HN

But it usually makes sense new models probably use somewhat same training pipelines and data.