Indexnusrefather/Erebus-RP-12B-Instruct-2608-v1
LoRA-style (fewer layers) vs google/gemma-3-12b-it. Indexnusrefather/Erebus-RP-12B-Instruct-2608-v1 matches part of google/gemma-3-12b-it (626 of its 1065 tensors; the rest of google/gemma-3-12b-it 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
4 derivatives checked
- mradermacher/Erebus-RP-12B-Instruct-2608-v1-i1-GGUF: Quantization
- mradermacher/Erebus-RP-12B-Instruct-2608-v1-GGUF: Quantization
- ailexleon/Erebus-RP-12B-Instruct-2608-v1-mlx-6Bit: Quantization
- ailexleon/Erebus-RP-12B-Instruct-2608-v1-mlx-4Bit: Quantization