BlossomsAI/Sailor2-8B-Chat-Uncensored
Abliterated vs sail/Sailor2-8B-Chat. Looks like abliteration (refusal removal): only the output projections (down_proj, embed_tokens, o_proj) changed, in 32 of 32 layers. Abliteration edits exactly these matrices to remove a "refusal direction"; everything else is untouched.
Tensors changed: Embeddings 1/1, Norms 0/65, MLP 32/96, Attention 32/224, lm_head 0/1
3 derivatives checked
- mradermacher/Sailor2-8B-Chat-Uncensored-i1-GGUF: Quantization
- mradermacher/Sailor2-8B-Chat-Uncensored-GGUF: Same weights, converted
- BlossomsAI/Sailor2-8B-Chat-Uncensored-GGUF: Quantization