Labonne's article says the second step is to "Calculate the mean difference between the activations of harmful and harmless instructions", which gives a vector representing the "refusal direction" for each layer, and the third is to "Normalize these vectors and evaluate them to select the single best 'refusal direction.'"

The exact words it rests on

"Mean difference : Calculate the mean difference between the activations of harmful and harmless instructions. This gives us a vector representing the "refusal direction" for each layer of the model. Selection : Normalize these vectors and evaluate them to select the single best "refusal direction.""

What the source said when we opened it, on 26 Oct 2026.

The source

Uncensor any LLM with abliteration
Hugging Face community article (Maxime Labonne) · 2024-06-13

Open the sourcearchived copy

Checked

Checked by the notis newsroom on 26 Oct 2026, against the source above.

In the story

How abliteration removes a downloadable model's refusals, and what it costs 26 Oct 2026