NLP/bert-utils

一行代码使用BERT生成句向量，BERT做文本分类、文本相似度计算

Go to file

joe 7ab44a43a6 添加代码		2019-01-29 18:31:51 +08:00
data	添加代码	2019-01-29 18:31:51 +08:00
__init__.py	添加代码	2019-01-29 18:31:51 +08:00
args.py	添加代码	2019-01-29 18:31:51 +08:00
extract_feature.py	添加代码	2019-01-29 18:31:51 +08:00
extract_features_or.py	添加代码	2019-01-29 18:31:51 +08:00
graph.py	添加代码	2019-01-29 18:31:51 +08:00
modeling.py	添加代码	2019-01-29 18:31:51 +08:00
optimization.py	添加代码	2019-01-29 18:31:51 +08:00
README.md	添加代码	2019-01-29 18:31:51 +08:00
requirements.txt	添加代码	2019-01-29 18:31:51 +08:00
similarity.py	添加代码	2019-01-29 18:31:51 +08:00
tokenization.py	添加代码	2019-01-29 18:31:51 +08:00

README.md

bert_feature

how to use Bert generate the sentence vector

1、download the model

model path: https://storage.googleapis.com/bert_models/2018_11_03/chinese_L-12_H-768_A-12.zip

2、Move the model in the same directory

3、init BertVector object and invokes the encode method, the param must be list

from bert.extrac_feature import BertVector
bv = BertVector()
bv.encode(['你好'])