Attention Sees, MLP Refuses: Tracing the Attention-to-MLP Safety Pathway in LLMs

cs.AI

Submitted: 2026-10-09

Updated: 2026-10-09

Comments: Preprint

Code: https://github.com/tatsu-lab/alpaca_eval

License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/

Sources

• Eliciting Latent Predictions from Transformers with the Tuned Lens
• How Post-Training Reshapes LLMs: A Mechanistic View on Knowledge, Truthfulness, Refusal, and Confidence
• The Llama 3 Herd of Models
• AlphaSteer: Learning Refusal Steering with Principled Null-Space Constraint
• Universal and Transferable Adversarial Attacks on Aligned Language Models

Related papers

• MAVEN-T: Reinforced Heterogeneous Distillation for Real-Time Multi-Agent Trajectory Prediction
• Model Discovery Agent: LLM-assisted Bayesian experiment design for data-efficient discovery of mechanistic world models
• Agents' Last Exam
• The Clinician's Veto: Navigating Trust, Liability, and Uncertainty in Autonomous AI Prescribing
• MindHelper: Closed-Loop Embodied Mental-State Reasoning for Precision Intervention
• Incumbent Advantage: Brand Bias and Cognitive Manipulation Dynamics in LLM Recommendation Systems