huihui-ai/Huihui-gemma-4-E2B-it-abliterated
Abliterated vs google/gemma-4-E2B-it. Looks like abliteration (refusal removal): only the output projections (down_proj, o_proj) changed, in 25 of 35 layers. Abliteration edits exactly these matrices to remove a "refusal direction"; everything else is untouched.
Tensors changed: MLP 25/585, Norms 0/453, other 0/240, Attention 25/727, Embeddings 0/6
5 derivatives checked
- mradermacher/Huihui-gemma-4-E2B-it-abliterated-GGUF: Same weights, converted (fewer layers)
- mradermacher/Huihui-gemma-4-E2B-it-abliterated-i1-GGUF: Quantization (fewer layers)
- EZCon/Huihui-gemma-4-E2B-it-abliterated-4bit-mlx: Quantization
- EZCon/Huihui-gemma-4-E2B-it-abliterated-4bit-g32-mxfp4-mixed_4_8-mlx: Quantization
- EZCon/Huihui-gemma-4-E2B-it-abliterated-8bit-mlx: Quantization
3 other repos hold byte-for-byte identical weight files.