
gradient-untangler
Local white-box gradient attacks for open-weight LLMs: GCG/PEZ suffix search, layer saliency, weight snapshots, and rank-1 suffix-to-delta fitting…
adversarial-attackai-securitymachine-learning+1

Local white-box gradient attacks for open-weight LLMs: GCG/PEZ suffix search, layer saliency, weight snapshots, and rank-1 suffix-to-delta fitting…