Lurkhunter / Chinese-Text-Classification

Chinese-Text-Classification Project including bert-classification, textCNN and so on.

Geek Repo:Geek Repo

Github PK Tool:Github PK Tool

Text Classification(pytorch)

本项目是关于一些文本分类任务的一些模型实现及笔记,持续更新.

数据集: [THUCNews](http://thuctc.thunlp.org/) ,本项目只选取了部分短文本数据集,已上传,data目录下。 数据:10个类
{'体育': 0, '娱乐': 1, '家居': 2, '房产': 3, '教育': 4, '时尚': 5, '时政': 6, '游戏': 7, '科技': 8, '财经': 9}

Result

model Accuracy Loss time(s)/epoch pretrain-model-download-link algorithm nodes
bert-base 0.9102 0.3382 137.7193 bert_chinese 预训练模型:从BERT到XLNet、RoBERTa、ALBERT
Albert-tiny 0.7961 0.6357 15.6404 albert_chinese 预训练模型:从BERT到XLNet、RoBERTa、ALBERT
macbert-base 0.9156 0.2811 137.4945 macbert_chinese wait to update
TextCNN 0.8773 0.6357 28.8052 None 搭一个TextCNN-文本分类利器

不同模型结果的详细结果报告、如何运行等,详见单个项目内README.md。

About

Chinese-Text-Classification Project including bert-classification, textCNN and so on.


Languages

Language:Python 100.0%