ALBERT for Chinese Text Classification

albert-zh finetune for tensorflow 2.x

基于tf2及albert_zh的中文情感分析任务

tensorflow 2.3
python 3.8
albert-tiny-zh

##label

0: negative
1: positive

注意调整FLAGS配置

python run_classifier.py

python predict_from_saved_model.py

example

"水果新鲜！发货快，服务好，京东物流顶呱呱，快递小哥服务好周到，送货上门，每次都热情满满，辛苦了，必须赞！",
"隔音差，加速有点不给力",
"最满意的是耗电量很低，百公里耗电才11.8度，这样算起来，出行成本比地铁的还低。"

output

cost time: 0.05300021171569824
['positive', 'negative', 'positive']

https://github.com/brightmart/albert_zh

https://github.com/kpe/bert-for-tf2

https://github.com/kamalkraj/ALBERT-TF2.0

Name		Name	Last commit message	Last commit date
Latest commit History 5 Commits
albert_tiny		albert_tiny
bert		bert
data		data
output		output
saved_model/1		saved_model/1
tf_record_dir		tf_record_dir
.gitignore		.gitignore
LICENSE		LICENSE
README.md		README.md
albert_config.py		albert_config.py
classifier_data_lib.py		classifier_data_lib.py
create_finetuning_data.py		create_finetuning_data.py
input_pipeline.py		input_pipeline.py
optimization.py		optimization.py
predict_from_saved_model.py		predict_from_saved_model.py
requirements.txt		requirements.txt
run_classifier.py		run_classifier.py
similarity.py		similarity.py
similarity_added.py		similarity_added.py
squad_lib.py		squad_lib.py
tokenization.py		tokenization.py