junafinity/Ornith-1.5-9B-uncensored
Abliterated vs ornith-ai/Ornith-1.5-9B. Looks like abliteration (refusal removal): only the output projections (down_proj, o_proj, out_proj) changed, in 6 of 32 layers. Abliteration edits exactly these matrices to remove a "refusal direction"; everything else is untouched.
Tensors changed: Embeddings 0/4, Norms 0/215, Attention 6/332, MLP 6/208, lm_head 0/1
8 derivatives checked
- mradermacher/Ornith-1.5-9B-uncensored-GGUF: Same weights, converted
- mradermacher/Ornith-1.5-9B-uncensored-i1-GGUF: Quantization
- N05off/Ornith-1.5-9B-uncensored-GGUF: Same weights, converted
- Lazybit666/Ornith-1.5-9B-uncensored-GGUF: Same weights, converted
- Sanheez/Ornith-1.5-9B-uncensored-GGUF: Same weights, converted
- cutetitki/Ornith-1.5-9B-uncensored-GGUF: Same weights, converted
- hsjdjejejeieieieie/Ornith-1.5-9B-uncensored-GGUF: Same weights, converted
- keXjos/Ornith-1.5-9B-uncensored-mlx-5Bit: Quantization