Efficient LLM Jailbreak via Adaptive Dense-to-sparse Constrained Optimization

Open in new window