Issue with embedding layer when serving a Tensorflow/Keras model with TF 2.0

I followed the step in one of the TF beginner tutorial to create a simple classification model. They are the following:

from __future__ import absolute_import, division, print_function, unicode_literals
import numpy as np
import pandas as pd
import tensorflow as tf
from tensorflow import feature_column
from tensorflow.keras import layers
from sklearn.model_selection import train_test_split

URL = 'https://storage.googleapis.com/applied-dl/heart.csv'
dataframe = pd.read_csv(URL)
dataframe.head()

train, test = train_test_split(dataframe, test_size=0.2)
train, val = train_test_split(train, test_size=0.2)

def df_to_dataset(dataframe, shuffle=True, batch_size=32):
  dataframe = dataframe.copy()
  labels = dataframe.pop('target')
  ds = tf.data.Dataset.from_tensor_slices((dict(dataframe), labels))
  if shuffle:
    ds = ds.shuffle(buffer_size=len(dataframe))
  ds = ds.batch(batch_size)
  return ds

batch_size = 5 # A small batch sized is used for demonstration purposes
train_ds = df_to_dataset(train, batch_size=batch_size)
val_ds = df_to_dataset(val, shuffle=False, batch_size=batch_size)
test_ds = df_to_dataset(test, shuffle=False, batch_size=batch_size)

feature_columns = []
for header in ['age', 'trestbps', 'chol', 'thalach', 'oldpeak', 'slope', 'ca']:
  feature_columns.append(feature_column.numeric_column(header))
thal_embedding = feature_column.embedding_column(thal, dimension=8)
feature_columns.append(thal_embedding)

feature_layer = tf.keras.layers.DenseFeatures(feature_columns)

batch_size = 32
train_ds = df_to_dataset(train, batch_size=batch_size)
val_ds = df_to_dataset(val, shuffle=False, batch_size=batch_size)
test_ds = df_to_dataset(test, shuffle=False, batch_size=batch_size)


model = tf.keras.Sequential([
  feature_layer,
  layers.Dense(128, activation='relu'),
  layers.Dense(128, activation='relu'),
  layers.Dense(1, activation='sigmoid')
])

model.compile(optimizer='adam',
              loss='binary_crossentropy',
              metrics=['accuracy'],
              run_eagerly=True)

model.fit(train_ds,
          validation_data=val_ds,
          epochs=5)

And I saved the model with:

model.save("model/", save_format='tf')

Then, I try to serve this model using this TF tutorial. I do the following:

docker pull tensorflow/serving
docker run -p 8501:8501 --mount type=bind,source=/path/to/model/,target=/models/model -e MODEL_NAME=mo

And I try to call the model this way:

curl -d '{"inputs": {"age": [0], "trestbps": [0], "chol": [0], "thalach": [0], "oldpeak": [0], "slope": [1], "ca": [0], "exang": [0], "restecg": [0], "fbs": [0], "cp": [0], "sex": [0], "thal": ["normal"], "target": [0] }}' -X POST http://localhost:8501/v1/models/model:predict

I get the following error:

{ "error": "indices = 1 is not in [0, 1)\n\t [[]]" }

It seems to be related to the embedding layer for the "thal" feature. But I have no idea what "indices = 1 is not in [0, 1)" means and why it happens.

When the error occurs, here is what the TF docker server logs:

2019-09-23 12:50:43.921721: W external/org_tensorflow/tensorflow/core/framework/op_kernel.cc:1502] OP_REQUIRES failed at lookup_table_op.cc:952 : Failed precondition: Table already initialized.

Any idea where the error comes from and how I could fix it?

Python version: 3.6

tensorflow version: 2.0.0-rc0

latest TensorFlow/serving (as of 20/09/2019)

from Issue with embedding layer when serving a Tensorflow/Keras model with TF 2.0