MMODELYST
People/Zhipeng Wang
PPL

Zhipeng Wang

Papers (1)
Beyond GRPO and On-Policy Distillation: An Empirical Sparse-to-Dense Reward Principle for Language-Model Post-Training