Hacker Newsnew | past | comments | ask | show | jobs | submitlogin

They're using specialized hardware to accelerate their development feedback loop. Without a doubt researchers and hackers will find ways to cut down model sizes and complexity, to run on consumer hardware, soon enough. Just use stable diffusion as an example: 4GB for the whole model. Even if text models are 16GB that'd be great.


We can't easily replicate it if the underlying algorithm isn't being disclosed. We'd need to rediscover whatever new tricks they used.




Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: