Yuyang Liu

Clipping Bottleneck: Stabilizing RLVR via Stochastic Recovery of Near-Boundary Signals

Shuo Yang, Jinda Lu, Chiyu Ma, et al.

cuGenOpt: A GPU-Accelerated General-Purpose Metaheuristic Framework for Combinatorial Optimization

Yuyang Liu

cuGenOpt is a GPU-accelerated software framework that solves complex combinatorial optimization problems (like logistics routing and scheduling) much faster than traditional methods. It balances three key goals: being general-purpose enough for many problem types, running efficiently on graphics processors, and remaining easy for users to apply—including a natural language interface powered by AI. Testing shows it outperforms standard solvers by orders of magnitude and can solve problems with hundreds of variables effectively.