AI.info

Yu Li

Explore Yu Li on AI.info.

• SafeEvolve: Harness-Policy Co-Evolution from Agent Experience for Safety Alignment
• StepGuard: Learning Step-Level Guardrails with Scalable Supervision and Safety-Utility Balancing
• On the Fragility of Self-Improving Agents: Variance, Task Order, and Underspecification
• SHE: Trajectory-driven Safety Harness Evolution for LLM Agents
• Frontier AI Risk Management Framework in Practice: A Risk Analysis Technical Report v1.5
• ETS: Energy-Guided Test-Time Scaling for Training-Free RL Alignment
• Exploring the Use of VLMs for Navigation Assistance for People with Blindness and Low Vision
• AgentDoG: A Diagnostic Guardrail Framework for AI Agent Safety and Security
• Agentic reinforcement learning empowers next-generation chemical language models for molecular design and synthesis
• CuMA: Aligning LLMs with Sparse Cultural Values via Demographic-Aware Mixture of Adapters
• Beyond Semantic Features: Pixel-level Mapping for Generalized AI-Generated Image Detection
• ArcGen: Generalizing Neural Backdoor Detection Across Diverse Architectures
• Investigating Data Pruning for Pretraining Biological Foundation Models at Scale
• From Static Structures to Ensembles: Studying and Harnessing Protein Structure Tokenization
• C$^3$TG: Conflict-aware, Composite, and Collaborative Controlled Text Generation