|
--- |
|
language: |
|
- zh |
|
|
|
license: apache-2.0 |
|
|
|
tags: |
|
- classification |
|
|
|
inference: false |
|
|
|
--- |
|
|
|
# Erlangshen-MacBERT-325M-TextMatch-Chinese |
|
|
|
- Github: [Fengshenbang-LM](https://github.com/IDEA-CCNL/Fengshenbang-LM) |
|
- Docs: [Fengshenbang-Docs](https://fengshenbang-doc.readthedocs.io/) |
|
|
|
## 简介 Brief Introduction |
|
|
|
110M参数的Topic Classification BERT (TCBert),。 |
|
|
|
The TCBert with 110M parameters is pre-trained for, not limited to, Chinese topic classification tasks. |
|
|
|
## 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 | 自然语言理解 NLU | 二郎神 Erlangshen | TCBert | 110M | Chinese | |
|
|
|
## 模型信息 Model Information |
|
|
|
|
|
为了提高模型在句子匹配上的效果,我们收集了大量句子匹配数据进行预训练,随后在FewCLUE的BUSTM任务进行微调,所有的训练均基于我们提出的UniMC框架。最终结果表明,3.25亿参数的模型通过我们的训练策略可以在句子匹配任务上超过1.3亿参数的大模型。 |
|
|
|
To improve the model performance on the topic classification task, we collected numerous topic classification datasets for pre-training based on general prompts. |
|
### 下游效果 Performance |
|
|
|
Stay tuned!!! |
|
|
|
## 使用 Usage |
|
|
|
```python |
|
from transformers import BertForMaskedLM, BertTokenizer |
|
import torch |
|
tokenizer=BertTokenizer.from_pretrained("IDEA-CCNL/Erlangshen-TCBert-110M-Classification-Chinese") |
|
model=BertForMaskedLM.from_pretrained('IDEA-CCNL/Erlangshen-TCBert-110M-Classification-Chinese') |
|
``` |
|
|
|
如果您在您的工作中使用了我们的模型,可以引用我们的[网站](https://github.com/IDEA-CCNL/Fengshenbang-LM/): |
|
|
|
You can also cite our [website](https://github.com/IDEA-CCNL/Fengshenbang-LM/): |
|
|
|
```text |
|
@misc{Fengshenbang-LM, |
|
title={Fengshenbang-LM}, |
|
author={IDEA-CCNL}, |
|
year={2021}, |
|
howpublished={\url{https://github.com/IDEA-CCNL/Fengshenbang-LM}}, |
|
} |
|
``` |