soob3123/amoral-gemma3-12B-v1
LoRA-style vs google/gemma-3-12b-it. Looks like a merged LoRA (or a fine-tune with frozen norms): only the weight matrices of down_proj, gate_proj, k_proj, o_proj, q_proj, up_proj, v_proj changed, in 48 of 48 layers; norms, biases and embeddings are untouched.
Tensors changed: Embeddings 0/4, Norms 0/400, MLP 144/252, Attention 192/408, other 0/1
5 derivatives checked
- bartowski/soob3123_amoral-gemma3-12B-GGUF: Same weights, converted (fewer layers)
- mradermacher/amoral-gemma3-12B-i1-GGUF: Quantization (fewer layers)
- mradermacher/amoral-gemma3-12B-GGUF: Quantization (fewer layers)
- DevQuasar-11/soob3123.amoral-gemma3-12B-GGUF: Same weights, converted (fewer layers)
- gghfez/amoral-gemma3-12B-vision: Identical copy
2 other repos hold byte-for-byte identical weight files.