ToastyPigeon/Gemma-3-Confetti-12B
LoRA-style (fewer layers) vs google/gemma-3-12b-pt. ToastyPigeon/Gemma-3-Confetti-12B matches part of google/gemma-3-12b-pt (626 of its 1065 tensors; the rest of google/gemma-3-12b-pt has no counterpart). On the shared part: 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 0/1, Norms 0/289, MLP 144/144, Attention 192/192
7 derivatives checked
- mradermacher/Gemma-3-Confetti-12B-i1-GGUF: Quantization
- ArtusDev/ToastyPigeon_Gemma-3-Confetti-12B-GGUF: Quantization
- mradermacher/Gemma-3-Confetti-12B-GGUF: Quantization
- ToastyPigeon/Gemma-3-Starshine-12B: Full fine-tune (fewer layers)
- ToastyPigeon/Gemma-3-Starshine-12B-Alt: Full fine-tune
- ToastyPigeon/glitterfetti-merge-A2: Full fine-tune
- ToastyPigeon/glitterfetti-merge-A3: Full fine-tune