PPL
Chengsong Huang
Papers (5)
You Only Need Minimal RLVR Training: Extrapolating LLMs via Rank-1 TrajectoriesProcess Rewards with Learned ReliabilityG-Zero: Self-Play for Open-Ended Generation from Zero DataLLMs Improving LLMs: Agentic Discovery for Test-Time ScalingNonsense Helps: Prompt Space Perturbation Broadens Reasoning Exploration