Guard Models Are Overconfident Where Base Models Are Uncertain
cs.LG, cs.AI
Submitted: 2026-09-29
Updated: 2026-09-29
Sources
- Jailbreaking Black Box Large Language Models in Twenty Queries
- Red Teaming Language Models to Reduce Harms: Methods, Scaling Behaviors, and Lessons Learned
- AEGIS: Online Adaptive AI Content Safety Moderation with Ensemble of LLM Experts
- The Llama 3 Herd of Models
- Llama Guard: LLM-based Input-Output Safeguard for Human-AI Conversations
- Mistral 7B
- Language Models (Mostly) Know What They Know
- Teaching Models to Express Their Uncertainty in Words
- AutoDAN-Turbo: A Lifelong Agent for Strategy Self-Exploration to Jailbreak LLMs
- Unlocking the Pre-Trained Model as a Dual-Alignment Calibrator for Post-Trained LLMs
- Trained on Tokens, Calibrated on Concepts: The Emergence of Semantic Calibration in LLMs
- GPT-4 Technical Report
- Guarding the Meaning: Self-Supervised Training for Semantic Robustness in Guard Models
- BaseCal: Unsupervised Confidence Calibration via Base Model Signals
- Gemma 2: Improving Open Language Models at a Practical Size
- Llama 2: Open Foundation and Fine-Tuned Chat Models
- Ethical and social risks of harm from Language Models
- TextGrad: Automatic "Differentiation" via Text
- ShieldGemma: Generative AI Content Moderation Based on Gemma
- Universal and Transferable Adversarial Attacks on Aligned Language Models
Related papers
- Polynomial-Augmented Neural Networks (PANNs) with Weak Orthogonality Constraints for Enhanced Function and PDE Approximation
- AIRL-S: Unifying Reinforcement Learning and Search-Based Test-Time Scaling via Adversarial Inverse Reinforcement Learning
- Transformers as Bayesian In-Context Experimenters: Smoothness-Adaptive Efficient ATE Estimation
- Convergence issues in Relational Concept Analysis based on AOC-posets
- Beliefs Beyond Posteriors: The Predictive Cost of ELBO Binding
- Understanding Diffusion Models via Ratio-Based Function Approximation with SignReLU Networks