步驟

• Step1：定義 @tf.function
• Step2：匯出 Save Model 自訂模組
• Step3：怎麼呼叫 TensorFlow Serving 的 RESTful API

Step1: 定義 @tf.function

class MLP(tf.keras.Model):
    ...

    @tf.function(input_signature=[tf.TensorSpec([None, 28, 28, 1], tf.float32)])
    def call(self, inputs):
    
    ...

Step2：匯出 Save Model 自訂模組

model = MLP()
tf.saved_model.save(model, "MLP/1", signatures={"call": model.call})

Step3：怎麼呼叫 TensorFlow Serving 的 RESTful API

import json
import requests
from zh.model.utils import MNISTLoader

data_loader = MNISTLoader()

data = json.dumps({
    "signature_name": "call",
    "instances": data_loader.test_data[0:10].tolist()
    })
headers = {"content-type": "application/json"}
json_response = requests.post(
    'http://localhost:8501/v1/models/MLP:predict',
    data=data, headers=headers)