News for hackers. Stuff that matters.

A comment on Show HN: 1-Bit Bonsai, the First Commercially Viable 1-Bit LLMs. Parent comment

Re:Show HN: 1-Bit Bonsai, the First Commercially Viable 1-Bi... by MarsIronPI on Tuesday March 31, 2026 @10:23PM (#47596028)
It's because they're natively trained with 1 bit, so it's not losing anything. Now, the question might be how they manage to get decent predictive performance with such little precision. That I don't know.
[ Parent | on HN ]
Re:Show HN: 1-Bit Bonsai, the First Commercially Viable 1-Bi... by syntaxpr on Wednesday April 01, 2026 @12:50AM (#47596949)
Not training. Transposing rows/columns of matrices to group 128 parameters with similar (shared) scale factor. Qwen-3 model.
[ Parent | on HN ]
Re:Show HN: 1-Bit Bonsai, the First Commercially Viable 1-Bi... by MarsIronPI on Wednesday April 01, 2026 @10:10AM (#47601144)
I'm not sure what you mean. Could you please elaborate?
[ Parent | on HN ]