bert-as-service 尝试

         肖涵博士，bert-as-service 作者。现为腾讯 AI Lab 高级科学家、德中人工智能协会主席。

启动server：

bert-serving-start -model_dir uncased_L-12_H-768_A-12 -num_worker=4
/home/zhongling/tensorflow1.4/lib/python3.5/site-packages/h5py/__init__.py:36: FutureWarning: Conversion of the second argument of issubdtype from `float` to `np.floating` is deprecated. In future, it will be treated as `np.float64 == np.dtype(float).type`.
  from ._conv import register_converters as _register_converters
usage: /home/zhongling/tensorflow1.4/bin/bert-serving-start -model_dir uncased_L-12_H-768_A-12 -num_worker=4
                 ARG   VALUE
__________________________________________________
           ckpt_name = bert_model.ckpt
         config_name = bert_config.json
                cors = *
                 cpu = False
          device_map = []
                fp16 = False
 gpu_memory_fraction = 0.5
       graph_tmp_dir = None
    http_max_connect = 10
           http_port = None
        mask_cls_sep = False
      max_batch_size = 256
         max_seq_len = 25
           model_dir = uncased_L-12_H-768_A-12
          num_worker = 4
       pooling_layer = [-2]
    pooling_strategy = REDUCE_MEAN
                port = 5555
            port_out = 5556
       prefetch_size = 10
 priority_batch_size = 16
     tuned_model_dir = None
             verbose = False
                 xla = False

I:VENTILATOR:[__i:__i: 63]:freeze, optimize and export graph, could take a while...
I:GRAPHOPT:[gra:opt: 52]:model config: uncased_L-12_H-768_A-12/bert_config.json
I:GRAPHOPT:[gra:opt: 55]:checkpoint: uncased_L-12_H-768_A-12/bert_model.ckpt
I:GRAPHOPT:[gra:opt: 59]:build graph...
I:GRAPHOPT:[gra:opt:128]:load parameters from checkpoint...
I:GRAPHOPT:[gra:opt:132]:optimize...

 

 

启动客户端：

>>> bc = BertClient("192.168.2.12")
 

>>> arr = ["women bag","phone case","Spring and summer lace bra wrapped chest anti cross chest sexy vest backing sling underwear (without PAD)","Factory Direct Fawn Decorations Bucket Bag","Fashion Women's Crossbody Bag Shoulder Bag With Cute Decoration","Flower Stud Earrings","Luxury Phone Case For iPhone"]
>>> for i in range(len(arr)):
...     print(i,arr[i])
...
0 women bag
1 phone case
2 Spring and summer lace bra wrapped chest anti cross chest sexy vest backing sling underwear (without PAD)
3 Factory Direct Fawn Decorations Bucket Bag
4 Fashion Women's Crossbody Bag Shoulder Bag With Cute Decoration
5 Flower Stud Earrings
6 Luxury Phone Case For iPhone
>>> vecs = bc.encode(arr)
>>> 1 - spatial.distance.cosine(vecs[1],vecs[2])
0.6363017559051514
>>> 1 - spatial.distance.cosine(vecs[1],vecs[3])
0.703334391117096
>>> 1 - spatial.distance.cosine(vecs[1],vecs[4])
0.6734864711761475
>>> 1 - spatial.distance.cosine(vecs[1],vecs[5])
0.7531731724739075
>>> 1 - spatial.distance.cosine(vecs[1],vecs[6])
0.7267242670059204