SHLTR/GLM-Z1-9B-0414-heretic
Abliterated vs zai-org/GLM-Z1-9B-0414. Looks like abliteration (refusal removal): only the output projections (down_proj) changed, in 33 of 40 layers. Abliteration edits exactly these matrices to remove a "refusal direction"; everything else is untouched.
Tensors changed: Embeddings 0/1, Norms 0/161, MLP 33/80, Attention 0/280, lm_head 0/1
2 derivatives checked
- mradermacher/GLM-Z1-9B-0414-heretic-i1-GGUF: Quantization
- mradermacher/GLM-Z1-9B-0414-heretic-GGUF: Same weights, converted