eggdog100/Qwable-v1-Qwen3.6-35B-A3B-abliterated
Abliterated vs lordx64/Qwable-v1. Looks like abliteration (refusal removal): only the output projections (down_proj, o_proj, out_proj) changed, in 21 of 40 layers. Abliteration edits exactly these matrices to remove a "refusal direction"; everything else is untouched.
Tensors changed: Embeddings 0/4, Attention 14/388, MLP 7/392, Norms 0/241, lm_head 0/1