3MPER0RR/gpt-oss-20b-3MPER0RR-abliterated
Abliterated vs wangzhang/gpt-oss-20b-abliterated. Looks like abliteration (refusal removal): only the output projections (o_proj) changed, in 7 of 24 layers. Abliteration edits exactly these matrices to remove a "refusal direction"; everything else is untouched.
Tensors changed: Embeddings 0/1, Norms 0/49, MLP 0/144, Attention 7/216, lm_head 0/1
2 derivatives checked
- mradermacher/gpt-oss-20b-3MPER0RR-abliterated-i1-GGUF: Quantization
- mradermacher/gpt-oss-20b-3MPER0RR-abliterated-GGUF: Quantization