The Era of 1-bit LLMs: ternary parameters for cost-effective computing(arxiv.org)
962 points by fgfm | 419 comments
arXiv paper introducing BitNet b1.58, a ternary-weight (-1, 0, 1) LLM matching full-precision Transformers while being far cheaper in latency, memory, and energy.




![Screenshot of youtube.com: Varda Capsule Reentry – Five Minutes from LEO to Earth [video]](https://hn.urlbox.com/shots/renders/2024/2/28/c22a257f-924b-4e50-bb33-8374bacfe2ef.jpg)




































![Screenshot of youtube.com: 2024 MIT Integration Bee – Finals [video]](https://hn.urlbox.com/shots/renders/2024/2/28/eb7a0471-35cf-467e-a4df-6628bea15109.jpg)

















