He He

Is it Thinking or Cheating? Detecting Implicit Reward Hacking by Measuring Reasoning Effort

Jailbreak Transferability Emerges from Shared Representations

Measuring LLM Novelty As The Frontier Of Original And High-Quality Output

LLM Watermark Evasion via Bias Inversion

Monitoring Decomposition Attacks with Lightweight Sequential Monitors

Unsupervised Elicitation of Language Models

Predicting Empirical AI Research Outcomes with Language Models

Adaptive Deployment of Untrusted LLMs Reduces Distributed Threats

Transformers Struggle to Learn to Search

Reasoning Models Know When They’re Right: Probing Hidden States for Self-Verification

Language Models Learn to Mislead Humans via RLHF

Hyperparameter Loss Surfaces Are Simple Near their Optima

An Asymptotic Theory of Random Search for Hyperparameters in Deep Learning