kerasを使った学習で出力の形が違うとエラーが出る

kerasを使って分類モデルを構築したい

kerasで分類モデルを作っているのですが、期待する出力シェイプとラベルデータのシェイプが違うというエラーが出てしまいます。色々と試しているのですが、解決法がわかりません。よろしくお願いします。

発生している問題・エラーメッセージ

INFO:plaidml:Opening device "metal_intel(r)_iris(tm)_plus_graphics.0"
_________________________________________________________________
Layer (type)                 Output Shape              Param #   
=================================================================
dense_1 (Dense)              (None, 10)                210       
=================================================================
Total params: 210
Trainable params: 210
Non-trainable params: 0
_________________________________________________________________
(None, 10)
Traceback (most recent call last):
  File "plaidml_mnist.py", line 143, in <module>
    model.train_on_batch(np.empty((5,20)), np.empty((5,10)))
  File "/Users/nakayamayasuaki/python_env/plaidml-venv/lib/python3.8/site-packages/keras/engine/training.py", line 1208, in train_on_batch
    x, y, sample_weights = self._standardize_user_data(
  File "/Users/nakayamayasuaki/python_env/plaidml-venv/lib/python3.8/site-packages/keras/engine/training.py", line 784, in _standardize_user_data
    y = standardize_input_data(
  File "/Users/nakayamayasuaki/python_env/plaidml-venv/lib/python3.8/site-packages/keras/engine/training_utils.py", line 134, in standardize_input_data
    raise ValueError(
ValueError: Error when checking target: expected dense_1 to have shape (1,) but got array with shape (10,)

該当のソースコード

plaidml.py
1#!/usr/bin/env python
2# coding: utf-8
3
4import plaidml.keras
5plaidml.keras.install_backend()
6import keras
7
8import numpy as np
9
10model = keras.Sequential([
11    keras.layers.Dense(units=10, activation='softmax', input_shape=(20,))
12])
13
14model.compile(loss='sparse_categorical_crossentropy',
15              optimizer='adam',
16              metrics=['accuracy'])
17
18model.summary()
19print(model.output_shape)
20model.train_on_batch(np.empty((5,20)), np.empty((5,10)))