Blackfrost-Research/GLM-5.3-F.U-AnthraClaud-Edition-BF16
Abliterated vs zai-org/GLM-5.3-BF16. Looks like abliteration (refusal removal): only the output projections (down_proj, o_proj) changed, in 48 of 79 layers. Abliteration edits exactly these matrices to remove a "refusal direction"; everything else is untouched.
Tensors changed: lm_head 0/1, Embeddings 0/1, Norms 0/364, MLP 429/2172, Attention 48/461, other 0/1