PPL
Tianyi Ma
Papers (5)
Why Semantic Entropy Fails: Geometry-Aware and Calibrated Uncertainty for Policy OptimizationHypergraph Pattern Machine: Compositional Tokenization for Higher-Order InteractionsSaaS-Bench: Can Computer-Use Agents Leverage Real-World SaaS to Solve Professional Workflows?PreScam: A Benchmark for Predicting Scam Progression from Early ConversationsOn the Safety of Graph Representation Learning