DavidAU/Gemma-3-4b-it-MAX-HORROR-Imatrix-GGUF
Quantization (fewer layers) vs google/gemma-3-4b-it. DavidAU/Gemma-3-4b-it-MAX-HORROR-Imatrix-GGUF matches part of google/gemma-3-4b-it (443 of its 883 tensors; the rest of google/gemma-3-4b-it has no counterpart). On the shared part: DavidAU/Gemma-3-4b-it-MAX-HORROR-Imatrix-GGUF is a GGUF quantization (Q8_0/BF16) of the same weights: 205 tensors match exactly and 238 within quantization error.
Tensors changed: Norms 0/205, MLP 0/102, Attention 0/136