LINEAiGE
AI HISTORY

PanGu-α Technical Paper Submission

The PanGu-α authors submitted the paper "PanGu-α: Large-scale Autoregressive Pretrained Chinese Language Models with Auto-parallel Computation" on April 26, 2021. The paper documents autoregressive language models with up to 200 billion parameters, developed with MindSpore and trained using 2048 Ascend 910 AI processors on 1.1 TB of Chinese pretraining data.

26 April 2021AI HISTORYVERIFIED

Historical classification

This verified record belongs to LINEAiGE's artificial-intelligence history layer beginning in 1956.

Why it matters

A primary technical-publication milestone documenting PanGu-α, a large-scale Chinese autoregressive pretrained language-model effort in 2021.

People, organizations & technologies

PEOPLE
Wei Zeng, Xiaozhe Ren, Teng Su, Hui Wang, Yi Liao, Zhiwei Wang, Xin Jiang, ZhenZhang Yang, Kaisheng Wang, Xiaoda Zhang, Chen Li, Ziyan Gong, Yifan Yao, Xinjing Huang, Jun Wang, Jianfeng Yu, Qi Guo, Yue Yu, Yan Zhang, Jin Wang, Hengtao Tao, Dasen Yan, Zexuan Yi, Fang Peng, Fangqing Jiang, Han Zhang, Lingfeng Deng, Yehong Zhang, Zhe Lin, Chao Zhang, Shaojie Zhang, Mingyue Guo, Shanzhi Gu, Gaojun Fan, Yaowei Wang, Xuefeng Jin, Qun Liu, Yonghong Tian

TECHNOLOGIES
PanGu-α, MindSpore, Ascend 910

Evidence & verification

STATE
verified

CONFIDENCE
high

LAST VERIFIED
2026-09-13T03:05:46+08:00

The authors' primary paper record supplies the April 26, 2021 submission date and directly documents PanGu-α models up to 200 billion parameters, MindSpore-based development, training on 2048 Ascend 910 processors, and 1.1 TB of Chinese pretraining data. Huawei's retrospective account separately states that the Pangu large model was officially released in April 2021. This candidate is intentionally limited to the paper-submission milestone; it does not convert Huawei's month-level release statement into an exact launch day and does not infer system-birth date, development geography, universal priority, ancestry, influence, causality, or relationship edges.

Sources

SOURCE 1 · PRIMARY · CLAIM-EVIDENCE

PanGu-α: Large-scale Autoregressive Pretrained Chinese Language Models with Auto-parallel Computation

PanGu-α authors / arXiv

PUBLISHED
2021-04-26

SOURCE 2 · PRIMARY · CORROBORATION

“盘古”开天记,AI落地时

Huawei

Trace the chronology

Record identity

LINEAiGE ID
pangu-alpha-paper-2021