YanLabs/Llama-3.3-8B-Instruct-MPOA
Abliterated vs shb777/Llama-3.3-8B-Instruct-128K. Looks like abliteration (refusal removal): only the output projections (down_proj, o_proj) changed, in 14 of 32 layers. Abliteration edits exactly these matrices to remove a "refusal direction"; everything else is untouched.
Tensors changed: Embeddings 0/1, Norms 0/65, MLP 14/96, Attention 10/128, lm_head 0/1
4 derivatives checked
- mradermacher/Llama-3.3-8B-Instruct-MPOA-i1-GGUF: Quantization
- mradermacher/Llama-3.3-8B-Instruct-MPOA-GGUF: Same weights, converted
- YanLabs/Llama-3.3-8B-Instruct-MPOA-GGUF: Same weights, converted
- SufficientPrune3897/Llama-3.3-8B-Character-Creator-V2: LoRA-style