tensorflow tfserving 部署记录

def get_mycustom_objects():
    # 模型在训练的时候会有自定义的一系列函数
    # 在从文件加载模型的时候，需要将其先前使用的一系列函数配件也传递给它
    my_objects = {
        'acc_top2': acc_top2,
        "metric_precision": metric_precision,
        "metric_recall": metric_recall,
        "metric_F1score": metric_F1score
    }
    from keras_bert import get_custom_objects
    custom_objects = get_custom_objects()
    custom_objects.update(my_objects)
    return custom_objects

def h5_to_pb(h5_model_path):
    """
    将h5模型转为pb格式
    :param h5_model_path:
    :return:
    """
    model = load_model(h5_model_path, compile=False, custom_objects=get_mycustom_objects())
    model.save('trained_model/1/', save_format='tf')
    print("转换结束")

MetaGraphDef with tag-set: 'serve' contains the following SignatureDefs:

signature_def['__saved_model_init_op']:
  The given SavedModel SignatureDef contains the following input(s):
  The given SavedModel SignatureDef contains the following output(s):
    outputs['__saved_model_init_op'] tensor_info:
        dtype: DT_INVALID
        shape: unknown_rank
        name: NoOp
  Method name is:

signature_def['serving_default']:
  The given SavedModel SignatureDef contains the following input(s):
    inputs['input_1'] tensor_info:
        dtype: DT_FLOAT
        shape: (-1, -1)
        name: serving_default_input_1:0
    inputs['input_2'] tensor_info:
        dtype: DT_FLOAT
        shape: (-1, -1)
        name: serving_default_input_2:0
  The given SavedModel SignatureDef contains the following output(s):
    outputs['dense_1'] tensor_info:
        dtype: DT_FLOAT
        shape: (-1, 4)
        name: StatefulPartitionedCall:0
  Method name is: tensorflow/serving/predict

import json
import requests, codecs
from keras_bert import Tokenizer
from keras.utils import to_categorical
import numpy as np

model_name ='xizang'
maxlen = 180
dict_path = r'bert/vocab.txt'

label2int = {
    "0": 0,
    "1": 1,
    "2": 2,
    "?": 3,
    "？": 3,
    "3": 3,
    "4": 1
}
numofcategories = len(set(label2int.values()))

# 重写tokenizer
class OurTokenizer(Tokenizer):
    def _tokenize(self, text):
        R = []
        for c in text:
            if c in self._token_dict:
                R.append(c)
            elif self._is_space(c):
                R.append('[unused1]')  # 用[unused1]来表示空格类字符
            else:
                R.append('[UNK]')  # 不在列表的字符用[UNK]表示   UNK是unknown的意思
        return R


def get_token_dict():
    """
    # 将词表中的字编号转换为字典
    :return: 返回自编码字典
    """
    token_dict = {}
    with codecs.open(dict_path, 'r', 'utf8') as reader:
        for line in reader:
            token = line.strip()
            token_dict[token] = len(token_dict)
    return token_dict


def text2input(text_list):
    tokenizer = OurTokenizer(get_token_dict())
    X1, X2, Y = [], [], []
    for text in text_list:
        text = text[:maxlen]
        x1, x2 = tokenizer.encode(first=text)
        X1.append(x1)
        X2.append(x2)
    return [X1, X2]


if __name__ == '__main__':
    textlist = ["芜湖起飞","大威天龙"]
    x = text2input(textlist)
    print(x)

    inputs = {}
    input_data = {
        # "signature_name": 'serving_default',
        # tfserving 有两种输入格式，一种是row/instances行格式，一种是column/inputs格式，本例使用第二种格式
        # column/inputs格式是，inputs为dict{}，每一个输入样本有两个输入input_1和input_2，将每一个样本的相同输入放在一个列表
        # row/instances格式是，instances为list[]，包含一个个instance，每一个instance是一个完整的样本输入，包含input_1和input_2
        # 参考https://www.tensorflow.org/tfx/serving/api_rest#request_format_2
        "inputs": {
            "input_1": x[0],  # input_1列表
            "input_2": x[1],  # input_2列表
        },
    }
    headers = {"content-type": "application/json"}
    data = json.dumps(input_data, indent=None)
    url = "http://192.168.70.163:8501/v1/models/" + model_name + ":predict"
    json_response = requests.post(url, data=data, headers=headers)
    # print(json_response.content)
    outputs = np.array(json.loads(json_response.text)['outputs'])
    print(outputs)
    y = [np.argmax(s) for s in outputs]
    print(y)