Quatfit/Quatfit-Mini-GGUF
Same weights, converted (fewer layers) vs Quatfit/Quatfit-Mini. Quatfit/Quatfit-Mini-GGUF matches part of Quatfit/Quatfit-Mini (623 of its 2076 tensors; the rest of Quatfit/Quatfit-Mini has no counterpart). On the shared part: Same weights stored in another dtype (F32 → F16/F32), plus 43 new tensors.
Tensors changed: Embeddings 0/2, Norms 0/278, MLP 0/126, other 0/85, Attention 0/132