| ▲ | Caltech Researchers Claim Compression of High-Fidelity AI Models(wsj.com) | |||||||
| 2 points by jonbaer 8 hours ago | 3 comments | ||||||||
| ▲ | WalterGR 6 hours ago | parent | next [-] | |||||||
105 comments: https://news.ycombinator.com/item?id=47593422 “Show HN: 1-Bit Bonsai, the First Commercially Viable 1-Bit LLMs” (prismml.com) | 227 points | 10 hours ago | ||||||||
| ▲ | grimm8080 8 hours ago | parent | prev [-] | |||||||
I need a subscription to read that. Though, I don't get why 1 bit LLMs are smaller, if you go from 8 bit to 1 bit, you'll need 8 times more nodes right? | ||||||||
| ||||||||