Machine Learning with TensorFlow 1.x

Performing a fine-tune on the model

In order to fine-tune the model, we need to know which one is the latest model and its corresponding checkpoint to restore weights and biases. Therefore, we call the /model endpoint to get the checkpoint name and a version number:

    def get_latest_model(url): 
    response = requests.get("%s/model" % url) 
    data = json.loads(response.text) 
    print(data) 
    return data["ckpt_name"], int(data["version"]) 

The response JSON should look like this:

    { 
     "ckpt_name": "2017-05-26_02-12-49",  
     "id": 10,  
     "link": "http://1.53.110.161:8181/pet-model/8.zip",  
     "name": "pet-model",  
     "version": 8 
    } 

Now, we will implement the code to fine-tune the model. Let's start with some parameters:

 # Server info URL = "http://localhost:5000" dest_api ...