First major Chinese-language LLM from Huawei. 200B dense parameters, trained on 2,048 Ascend 910 processors using MindSpore.

Outputs 2

PanGu-alpha

model
Architecture DENSE
Parameters 200B

PanGu-alpha: Large-scale Autoregressive Pretrained Chinese Language Models

paper

arXiv: 2104.12369

nlpopen-weight

Related