PPL
Hao Jiang
Papers (4)
Recovering Policy-Induced Errors: Benchmarking and Trajectory Synthesis for Robust GUI AgentsLong Live The Balance: Information Bottleneck Driven Tree-based Policy OptimizationThink When Needed: Adaptive Reasoning-Driven Multimodal Embeddings with a Dual-LoRA ArchitectureThe Extrapolation Cliff in On-Policy Distillation of Near-Deterministic Structured Outputs