llmfan46/gemma-4-26B-A4B-it-uncensored-heretic
Abliterated vs google/gemma-4-26B-A4B-it. Looks like abliteration (refusal removal): only the output projections (o_proj) changed, in 29 of 30 layers. Abliteration edits exactly these matrices to remove a "refusal direction"; everything else is untouched.
Tensors changed: Embeddings 0/4, MLP 0/231, Norms 0/433, other 0/122, Attention 29/223
11 derivatives checked
- llmfan46/gemma-4-26B-A4B-it-uncensored-heretic-GGUF: Same weights, converted
- mradermacher/gemma-4-26B-A4B-it-uncensored-heretic-i1-GGUF: Quantization
- noctrex/gemma-4-26B-A4B-it-uncensored-heretic-MXFP4_MOE-GGUF: Same weights, converted
- mradermacher/gemma-4-26B-A4B-it-uncensored-heretic-GGUF: Quantization
- Vastopian/gemma-4-26B-A4B-it-heretic-GGUF: Quantization
- ailexleon/gemma-4-26B-A4B-it-uncensored-heretic-mlx-vlm-6Bit: Quantization
- ailexleon/gemma-4-26B-A4B-it-uncensored-heretic-mlx-vlm-8Bit: Quantization
- Nestleshuffle/gemma-4-26B-A4B-it-uncensored-heretic-mlx-8Bit: Quantization
- jacob42/gemma-4-26B-A4B-it-uncensored-heretic-GGUF: Quantization
- Arue-2026/gemma-4-26B-A4B-it-uncensored-heretic-GGUF: Same weights, converted
- McG-221/gemma-4-26B-A4B-it-uncensored-heretic-9bit: Quantization