Tomorrow
Academia

Beijing, China

Peking University

Peking supplies much of the open-source Chinese model ecosystem's tooling — training frameworks, alignment datasets and video generation research.

AIChipsOfficial site
2 key findings0 timeline milestones1 sources to follow

Key findings

  1. 2023

    Safe RLHF and the BeaverTails dataset

    PKU-Alignment

    Separated helpfulness and harmlessness into distinct reward and cost models with a constrained optimisation objective.

    Why it matters here · The alignment recipe behind several open-weight models in the China cluster of the tracker.

  2. 2024

    Open video diffusion replication

    PKU-Yuan lab

    Reproduced spatiotemporal diffusion transformer video generation as fully open code and weights.

    Why it matters here · Why open video models reached the catalog within a year of the closed ones.

What to follow

PKU-Alignment

Community · Continuous

Safe RLHF datasets and alignment training code widely reused by open-weight Chinese models.

How to cite this page

Free to cite and reuse under CC BY 4.0. Permalink: https://tomorrow.aliensquad.ai/academia/peking

APA
Tomorrow. (2026). Peking University — key findings and research feeds [Research tracker entry]. AlienSquad. Retrieved 2026-09-13, from https://tomorrow.aliensquad.ai/academia/peking
BibTeX
@misc{tomorrow-academia-peking,
  author       = {{Tomorrow}},
  title        = {Peking University — key findings and research feeds},
  year         = {2026},
  publisher    = {AlienSquad},
  howpublished = {\url{https://tomorrow.aliensquad.ai/academia/peking}},
  note         = {Accessed: 2026-09-13}
}