llmfan46/GLM-Z1-32B-0414-uncensored-heretic-v1
Abliterated vs zai-org/GLM-Z1-32B-0414. Looks like abliteration (refusal removal): only the output projections (o_proj) changed, in 14 of 61 layers. Abliteration edits exactly these matrices to remove a "refusal direction"; everything else is untouched.
Tensors changed: Embeddings 0/1, Norms 0/245, MLP 0/122, Attention 14/244, lm_head 0/1
3 derivatives checked
- mradermacher/GLM-Z1-32B-0414-heretic-v1-i1-GGUF: Quantization
- mradermacher/GLM-Z1-32B-0414-heretic-v1-GGUF: Quantization
- llmfan46/GLM-Z1-32B-0414-uncensored-heretic-v1-GGUF: Same weights, converted