safwanahmadkhan/gemma-4-12B-it-DFlash-8bit-mlx-fp16
Same weights, converted vs z-lab/gemma4-12B-it-DFlash. Same weights stored in another dtype (BF16 → F16).
Tensors changed: other 0/1, Norms 0/22, MLP 0/15, Attention 0/20
Same weights, converted vs z-lab/gemma4-12B-it-DFlash. Same weights stored in another dtype (BF16 → F16).
Tensors changed: other 0/1, Norms 0/22, MLP 0/15, Attention 0/20