phruut/

Gemma-4-Gembrain-31B-gguf

i tried to make quants using Unsloth XL and Bartowski L recipes.
the XXL quant was made by comparing layers across the Unsloth, Bartowski, and mradermacher Q4_K_XL/L/M versions, and then keeping the highest quant per layer.
i didnt notice any issues after testing, but this is my first time making quants so there may still be issues.

llama.cpp version: b9244
imatrix from: https://huggingface.co/mradermacher/Gemma-4-Gembrain-31B-i1-GGUF