Shaofei Cai

TOWARD MEMORY-AIDED WORLD MODELS: BENCHMARKING VIA SPATIAL CONSISTENCY

UniCode: A Framework for Generating High-Quality Competitive Coding Problems

CUARewardBench: Benchmark for Evaluating Reward Models on Computer-using Agent Trajectories

Learn the Ropes, Then Trust the Wins: Self-imitation with Progressive Exploration for Agentic Reinforcement Learning

GROOT-2: Weakly Supervised Multimodal Instruction Following Agents

Optimizing Latent Goal by Learning from Trajectory Preference

Training Open-ended Policies to follow Video-prompt Instructions with Reinforcement Learning