← People

Yi Dong

Source-listed: Principal Research Scientist · NVIDIA

  • Language models
  • Reinforcement learning
  • Efficient ML

Selected work

5
  1. Nemotron-4 340B Technical ReportAuthor · 2024
  2. SteerLM: Attribute Conditioned SFT as an (User-Steerable) Alternative to RLHFContributor · 2023
  3. NeMo-Aligner: Scalable Toolkit for Efficient Model AlignmentContributor · 2024
  4. HelpSteer2: Open-source dataset for training top-performing reward modelsContributor · 2024
  5. ProRL: Prolonged Reinforcement Learning Expands Reasoning Boundaries in Large Language ModelsContributor · 2025