Brigham-Young-University/GLM-4.7-Flash-Ilograph
Abliterated vs zai-org/GLM-4.7-Flash. Looks like abliteration (refusal removal): only the output projections (o_proj) changed, in 47 of 47 layers. Abliteration edits exactly these matrices to remove a "refusal direction"; everything else is untouched.
Tensors changed: Embeddings 0/1, Norms 0/189, MLP 0/2504, Attention 47/235, lm_head 0/1
2 derivatives checked
- mradermacher/GLM-4.7-Flash-Ilograph-i1-GGUF: Quantization
- mradermacher/GLM-4.7-Flash-Ilograph-GGUF: Quantization