huihui-ai/Huihui-LFM2-8B-A1B-abliterated
Abliterated vs LiquidAI/LFM2-8B-A1B. Looks like abliteration (refusal removal): only the output projections (out_proj, w2) changed, in 24 of 24 layers. Abliteration edits exactly these matrices to remove a "refusal direction"; everything else is untouched.
Tensors changed: MLP 706/2162, Embeddings 0/2, other 18/54, Norms 0/60, Attention 6/24
3 derivatives checked
- mradermacher/Huihui-LFM2-8B-A1B-abliterated-i1-GGUF: Quantization
- mradermacher/Huihui-LFM2-8B-A1B-abliterated-GGUF: Same weights, converted
- Anotos/LFM2-8B-A1B-Abliterated-MLX-4bit: Quantization
1 other repos hold byte-for-byte identical weight files.