bert_serving 获取张量

安装：

pip install bert-serving-server  # server
pip install bert-serving-client  # client, independent of `bert-serving-server`
 

启动server

/data/apps/opt/anaconda3/bin/python /home/nlp/.local/bin/bert-serving-start -model_dir /data/nlp/pretrained_models/chinese_wwm_ext_L-12_H-768_A-12/ -num_worker=2 -port=5555 -port_out=5556

调用

from bert_serving.client import BertClient
bc = BertClient(ip='10.111.36.141')
test=bc.encode(['你好','bert'])
print(test)

输出

[[ 0.910996    0.09173454  0.31665277 ...  0.34575224  0.2235426
   0.26009858]
 [-0.4473213  -0.16672006 -0.03753631 ...  0.15099688 -0.52282304
  -0.24906637]]

------------- 分割线 -------------------

参数详解

参数	说明
-model_dir	预训练模型的路径
-num_worker	线程数，表示同时可以处理多少个并发请求
-pooling_strategy	默认是句向量；-pooling_strategy NONE是词向量
-max_seq_len	句子最大长度