AI
Bonsai 2 27B: How PrismML Shrinks a 27B Model to 5.9 GB
A 27B model that fits in about 6 GB sounds like a trick. It is ternary weights, a shared scale and a rotation. Here is how it works, what PrismML's own tests show, and where the loss hides.
Tensor · 2h ago · 8 min