ChiKoi7/GPT-5-Distill-llama3.2-3B-Instruct-Heretic-GGUF
Abliterated vs Jackrong/GPT-5-Distill-llama3.2-3B-Instruct. ChiKoi7/GPT-5-Distill-llama3.2-3B-Instruct-Heretic-GGUF is a GGUF (F16) of different weights. Looks like abliteration (refusal removal): only the output projections (down_proj, o_proj) changed, in 25 of 28 layers. Abliteration edits exactly these matrices to remove a "refusal direction"; everything else is untouched.
Tensors changed: Embeddings 0/1, Norms 0/57, MLP 23/84, Attention 25/112
1 derivatives checked
- blascotobasco/L3.2-Ascendant-Prime-16E-A6B: MoE built from dense models