PPL
Yue Huang
Papers (7)
Converted, Not Equivalent: Benchmarking Codebase Conversion via Observational EquivalenceAgentTrap: Measuring Runtime Trust Failures in Third-Party Agent SkillsVisual Aesthetic Benchmark: Can Frontier Models Judge Beauty?Prompt-Activation Duality: Improving Activation Steering via Attention-Level InterventionsSkillGen: Verified Inference-Time Agent Skill SynthesisNARRA-Gym for Evaluating Interactive Narrative AgentsCan Agents Price a Reaction? Evaluating LLMs on Chemical Cost Reasoning