mlabonne/gemma-3-27b-it-abliterated
Abliterated vs google/gemma-3-27b-it. Looks like abliteration (refusal removal): only the output projections (down_proj, o_proj) changed, in 62 of 62 layers. Abliteration edits exactly these matrices to remove a "refusal direction"; everything else is untouched.
Tensors changed: Embeddings 0/4, Norms 0/484, MLP 62/294, Attention 62/464, other 0/1
7 derivatives checked
- bartowski/mlabonne_gemma-3-27b-it-abliterated-GGUF: Same weights, converted (fewer layers)
- mradermacher/gemma-3-27b-it-abliterated-i1-GGUF: Quantization (fewer layers)
- mradermacher/gemma-3-27b-it-abliterated-GGUF: Quantization (fewer layers)
- danielmnd/mlabonne_gemma-3-27b-it-abliterated-GGUF: Same weights, converted (fewer layers)
- youkoutenhouin/gemma-3-27b-it-abliterated-GGUF: Quantization (fewer layers)
- summykai/gemma3-27b-abliterated-dpo: Identical copy
- mergekit-community/mergekit-model_stock-prczfmj: Abliterated
4 other repos hold byte-for-byte identical weight files.