alt Hacker Newssomething I have been wondering about is doing regressive layer specific quantization based on large test sets. ie reduce very specifically layers that don't improve general quality.
Replies
woadwarrior01 • yesterday at 10:32 PM
This is a very well established idea. It's called dynamic quantization. Vary the quantization bit-width (or skip quantization altogether) on a layer by layer basis, using a calibration dataset.
EvoPress is the first time that comes to my mind, when I think of dynamic quantization.
qskousen • today at 2:06 AM
I've experimented with this with diffusion models with a safetensors - gguf tool I wrote. even with relatively few sample images (~10k, still enough to keep my 3090 spinning for days straight) the benefits are quite noticeable - a smaller file with overall better results.
This is a thing! For example, https://arxiv.org/abs/2511.06516