PinoCookie/LFM2.5-2.6B-Abliterated
Abliterated vs LiquidAI/LFM2.5-2.6B. Looks like abliteration (refusal removal): only the output projections (out_proj, w2) changed, in 27 of 30 layers. Abliteration edits exactly these matrices to remove a "refusal direction"; everything else is untouched.
Tensors changed: Embeddings 0/2, other 20/66, MLP 25/90, Norms 0/76, Attention 7/32
2 derivatives checked
- mradermacher/LFM2.5-2.6B-Abliterated-i1-GGUF: Quantization
- mradermacher/LFM2.5-2.6B-Abliterated-GGUF: Same weights, converted