How do I export an eager execution model?

Question

Having completed my model, I now wish to export and deploy it, following this tutorial on TensorFlow's website. However, there is no indication of how to do this in eager execution, where I am unable to provide a session or graph to builder.add_meta_graph_and_variables().

Is this a case where my code needs to be eager and graph compatible, or where I need to save my model, import it to a session, and export it from there?

Answer 1

TF 2.0 Alpha supports exporting/saving an Eager Execution Model (Default is Eager Execution in 2.0). A SavedModel contains a complete TensorFlow program, including weights and computation.

Sample code for the same is shown below:

!pip install -q tensorflow==2.0.0-alpha0
import tensorflow as tf

pretrained_model = tf.keras.applications.MobileNet()
tf.saved_model.save(pretrained_model, "/tmp/mobilenet/1/")

#Loading the saved model
loaded = tf.saved_model.load("/tmp/mobilenet/1/")
infer = loaded.signatures["serving_default"]

Serving the Model:

nohup tensorflow_model_server \
  --rest_api_port=8501 \
  --model_name=mobilenet \
  --model_base_path="/tmp/mobilenet" >server.log 2>&1

#Sending the Request for Inference

!pip install -q requests
import json
import numpy
import requests
data = json.dumps({"signature_name": "serving_default",
                   "instances": x.tolist()})
headers = {"content-type": "application/json"}
json_response = requests.post('http://localhost:8501/v1/models/mobilenet:predict',
                              data=data, headers=headers)
predictions = numpy.array(json.loads(json_response.text)["predictions"])