Hao Zhou

PEGRL: Improving Machine Translation by Post-Editing Guided Reinforcement Learning

A Data-Efficient Path to Multilingual LLMs: Language Expansion via Post-training PARAM$Δ$ Integration into Upcycled MoE