MuXodious/Rocinante-X-12B-v1-absolute-heresy

Abliterated vs TheDrummer/Rocinante-X-12B-v1. Looks like abliteration (refusal removal): only the output projections (down_proj, o_proj) changed, in 31 of 40 layers. Abliteration edits exactly these matrices to remove a "refusal direction"; everything else is untouched.

Tensors changed: Embeddings 0/1, Norms 0/81, MLP 10/120, Attention 31/160, lm_head 0/1

4 derivatives checked

On the Hub · Downloads on Model Pulse