Dawn Song

8 papers UC Berkeley

• A Sustainable AI Economy Needs Data Deals That Work for Generators
• BountyBench: Dollar Impact of AI Agent Attackers and Defenders on Real-World Cybersecurity Systems
• OMEGA: Can LLMs Reason Outside the Box in Math? Evaluating Exploratory, Compositional, and Transformative Generalization
• OVERT: A Benchmark for Over-Refusal Evaluation on Text-to-Image Models
• SECODEPLT: A Unified Benchmark for Evaluating the Security Risks and Capabilities of Code GenAI
• Scalable Best-of-N Selection for Large Language Models via Self-Certainty
• VMDT: Decoding the Trustworthiness of Video Foundation Models
• Why and How LLMs Hallucinate: Connecting the Dots with Subsequence Associations