Cannae-AI/Gemini-3.1-pro-Gemma-4-E4B-Distill-gguf
Quantization (fewer layers) vs Cannae-AI/Gemini-3.1-pro-Gemma-4-E4B-Distill. Cannae-AI/Gemini-3.1-pro-Gemma-4-E4B-Distill-gguf matches part of Cannae-AI/Gemini-3.1-pro-Gemma-4-E4B-Distill (677 of its 2130 tensors; the rest of Cannae-AI/Gemini-3.1-pro-Gemma-4-E4B-Distill has no counterpart). On the shared part: Cannae-AI/Gemini-3.1-pro-Gemma-4-E4B-Distill-gguf is a GGUF quantization (Q8_0) of the same weights: 297 tensors match exactly and 380 within quantization error.
Tensors changed: Embeddings 0/2, Norms 0/296, MLP 0/126, other 0/85, Attention 0/168