annelo/GLM-5.2-FP8-Uncensored-Colibri-Int4
Abliterated vs jlnsrk/GLM-5.2-colibri-int4. Looks like abliteration (refusal removal): only the output projections (o_proj) changed, in 2 of 3 layers. Abliteration edits exactly these matrices to remove a "refusal direction"; everything else is untouched.
Tensors changed: lm_head 0/2, Embeddings 0/2, Norms 0/9, MLP 0/132, Attention 4/20