CoolFace
Modelpublic

IDEA-CCNL/Randeng-MegatronT5-770M

sourceHugging Faceapache-2.0updated 3y agoView on Hugging Face
7likes34downloads
Model Card

Randeng-MegatronT5-770M

简介 Brief Introduction

善于处理NLT任务，中文版的T5-large。

Good at solving NLT tasks, Chinese T5-large.

模型分类 Model Taxonomy

需求 Demand任务 Task系列 Series模型 Model参数 Parameter额外 Extra
通用 General自然语言转换 NLT燃灯 RandengMegatronT5770M中文-Chinese

模型信息 Model Information

为了得到一个大规模的中文版的T5，我们使用了Megatron-LM的方法和悟道语料库(180G版本)用于预训练。具体地，我们在预训练阶段中使用了Megatron-LM 大概花费了16张A100约14天。

To get a large-scale Chinese T5, we use of Megatron-LM and WuDao Corpora (180 GB version) for pre-training. Specifically, in the pre-training phase which cost about 14 days with 16 A100 GPUs.

使用 Usage

因为transformers库中是没有Randeng-MegatronT5-770M相关的模型结构的，所以你可以在我们的Fengshenbang-LM中找到并且运行代码。

Since there is no structure of Randeng-MegatronT5-770M in transformers library, you can find the structure of Randeng-MegatronT5-770M and run the codes in Fengshenbang-LM.

shell
 git clone https://github.com/IDEA-CCNL/Fengshenbang-LM.git

加载模型 Loading Models

python
from fengshen import T5ForConditionalGeneration
from fengshen import T5Config
from fengshen import T5Tokenizer

tokenizer = T5Tokenizer.from_pretrained('IDEA-CCNL/Randeng-MegatronT5-770M')
config = T5Config.from_pretrained('IDEA-CCNL/Randeng-MegatronT5-770M')
model = T5ForConditionalGeneration.from_pretrained('IDEA-CCNL/Randeng-MegatronT5-770M')

引用 Citation

如果您在您的工作中使用了我们的模型，可以引用我们的论文：

If you are using the resource for your work, please cite the our paper:

text
@article{fengshenbang,
  author    = {Jiaxing Zhang and Ruyi Gan and Junjie Wang and Yuxiang Zhang and Lin Zhang and Ping Yang and Xinyu Gao and Ziwei Wu and Xiaoqun Dong and Junqing He and Jianheng Zhuo and Qi Yang and Yongfeng Huang and Xiayu Li and Yanghan Wu and Junyu Lu and Xinyu Zhu and Weifeng Chen and Ting Han and Kunhao Pan and Rui Wang and Hao Wang and Xiaojun Wu and Zhongshen Zeng and Chongpei Chen},
  title     = {Fengshenbang 1.0: Being the Foundation of Chinese Cognitive Intelligence},
  journal   = {CoRR},
  volume    = {abs/2209.02970},
  year      = {2022}
}

也可以引用我们的网站:

You can also cite our website:

text
@misc{Fengshenbang-LM,
  title={Fengshenbang-LM},
  author={IDEA-CCNL},
  year={2021},
  howpublished={\url{https://github.com/IDEA-CCNL/Fengshenbang-LM}},
}