rank-one-safety-alignment

Apply rank-one weight modifications to amplify model safety via residual stream steering, requiring no fine-tuning and preserving utility on standard benchmarks

Independent GitHub snapshot · Automated classification · Original documentation governs