https://github.com/NLPScott/bert-Chinese-classification-task

bert中文分类实践
https://github.com/NLPScott/bert-Chinese-classification-task

Last synced: 6 days ago
JSON representation

bert中文分类实践

• Host: GitHub
• URL: https://github.com/NLPScott/bert-Chinese-classification-task
• Owner: NLPScott
• Created: 2018-11-26T10:21:25.000Z (almost 6 years ago)
• Default Branch: master
• Last Pushed: 2018-12-11T07:10:49.000Z (almost 6 years ago)
• Last Synced: 2024-08-02T08:09:55.706Z (3 months ago)
• Language: Python
• Size: 454 KB
• Stars: 733
• Watchers: 30
• Forks: 223
• Open Issues: 19
• Metadata Files:
  • Readme: README.md

Awesome Lists containing this project

• awesome-bert - NLPScott/bert-Chinese-classification-task
• awesome-transformer-nlp - NLPScott/bert-Chinese-classification-task - BERT Chinese classification practice. (Tasks / Classification)

README

        
# bert-Chinese-classification-task

bert中文分类实践

在run_classifier_word.py中添加NewsProcessor，即新闻的预处理读入部分 \

在main方法中添加news类型数据处理label \

 processors = { \

        "cola": ColaProcessor,\

        "mnli": MnliProcessor,\

        "mrpc": MrpcProcessor,\

        "news": NewsProcessor,\

    }

    

download_glue_data.py 提供glue_data下面其他的bert论文公测glue数据下载

data目录下是news数据的样例

export GLUE_DIR=/search/odin/bert/extract_code/glue_data \

export BERT_BASE_DIR=/search/odin/bert/chinese_L-12_H-768_A-12/ \

export BERT_PYTORCH_DIR=/search/odin/bert/chinese_L-12_H-768_A-12/

python run_classifier_word.py \

  --task_name NEWS \

  --do_train \

  --do_eval \

  --data_dir $GLUE_DIR/NewsAll/ \

  --vocab_file $BERT_BASE_DIR/vocab.txt \

  --bert_config_file $BERT_BASE_DIR/bert_config.json \

  --init_checkpoint $BERT_PYTORCH_DIR/pytorch_model.bin \

  --max_seq_length 256 \

  --train_batch_size 32 \

  --learning_rate 2e-5 \

  --num_train_epochs 3.0 \

  --output_dir ./newsAll_output/ \

  --local_rank 3

  

  中文分类任务实践

实验中对中文34个topic进行实践（包括：时政，娱乐，体育等），在对run_classifier.py代码中的预处理环节需要加入NewsProcessor模块，及类似于MrpcProcessor，但是需要对中文的编码进行适当修改，训练数据与测试数据按照4:1进行切割，数据量约80万，单卡GPU资源，训练时间18小时，acc为92.8%

eval_accuracy = 0.9281581998809113

eval_loss = 0.2222444740207354

global_step = 59826

loss = 0.14488934577978746