SC117/gemma-4-31B-it-heretic-QAT-For-Edge-16G-GGUF
Same weights, converted (fewer layers) vs coder3101/gemma-4-31B-it-qat-q4_0-unquantized-heretic. SC117/gemma-4-31B-it-heretic-QAT-For-Edge-16G-GGUF matches part of coder3101/gemma-4-31B-it-qat-q4_0-unquantized-heretic (772 of its 1188 tensors; the rest of coder3101/gemma-4-31B-it-qat-q4_0-unquantized-heretic has no counterpart). On the shared part: Same weights stored in another dtype (BF16 → F16/F32), plus 61 new tensors.
Tensors changed: Embeddings 0/1, Norms 0/361, MLP 0/180, Attention 0/230