venkycs/Mistral-Small-24B-Instruct-2501-Abliterated

Abliterated vs mistralai/Mistral-Small-24B-Instruct-2501. Looks like abliteration (refusal removal): only the output projections (down_proj, o_proj) changed, in 39 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 39/120, Attention 39/160, lm_head 0/1

1 derivatives checked

On the Hub · Downloads on Model Pulse