InvalidArgumentError：[0，0]中的预期大小为[0]，但得到1

Question

我正在尝试使用分类交叉熵损失来训练神经网络，但是我得到了

InvalidArgumentError:  Expected size[0] in [0, 0], but got 1
 [[node loss_4/dense_18_loss/categorical_crossentropy/softmax_cross_entropy_with_logits/Slice_1 
(defined at C:...\keras\backend\tensorflow_backend.py:3009) ]] [Op:__inference_keras_scratch_graph_2929]

Function call stack:
keras_scratch_graph

我不明白它在告诉我什么。是说它期望的标量值为0到0？

这是我的代码

model = Sequential()
model.add(layers.Dense(32, 
                        activation='relu',
                        input_shape=self.obsSpace))
model.add(layers.Dense(32, activation='relu'))
model.add(layers.Dense(self.actionSpace, 
                           activation='softmax'))

model.compile(loss="categorical_crossentropy",
                      optimizer=tf.keras.optimizers.Adam())

model.net.train_on_batch(batchStates, batchQVals)

batchStates为

tf.Tensor(
[[-4.16258150e-02  2.25938538e-01  4.36936698e-02 -2.81657789e-01]
 [-3.71070442e-02  4.20410887e-01  3.80605141e-02 -5.60246049e-01]
 [-2.86988265e-02  2.24775998e-01  2.68555931e-02 -2.55819147e-01]
 [-2.42033065e-02  4.19504426e-01  2.17392101e-02 -5.39911869e-01]
 [-1.58132180e-02  2.24083740e-01  1.09409728e-02 -2.40459278e-01]
 [-1.13315432e-02  4.19047703e-01  6.13178719e-03 -5.29671139e-01]
 [-2.95058915e-03  2.23840031e-01 -4.46163558e-03 -2.35062400e-01]
 [ 1.52621147e-03  2.87821088e-02 -9.16288359e-03  5.62098541e-02]
 [ 2.10185365e-03  2.24034234e-01 -8.03868651e-03 -2.39349889e-01]
 [ 6.58253833e-03  2.90280372e-02 -1.28256843e-02  5.07866067e-02]
 [ 7.16309907e-03  2.24331524e-01 -1.18099522e-02 -2.45915177e-01]
 [ 1.16497296e-02  4.19620142e-01 -1.67282557e-02 -5.42299721e-01]
 [ 2.00421324e-02  2.24737244e-01 -2.75742501e-02 -2.54934152e-01]
 [ 2.45368773e-02  4.20241828e-01 -3.26729332e-02 -5.56185350e-01]
 [ 3.29417139e-02  6.15806894e-01 -4.37966402e-02 -8.58980519e-01]
 [ 4.52578517e-02  8.11497186e-01 -6.09762505e-02 -1.16510658e+00]
 [ 6.14877955e-02  6.17219641e-01 -8.42783820e-02 -8.92147760e-01]
 [ 7.38321883e-02  8.13377481e-01 -1.02121337e-01 -1.21008870e+00]
 [ 9.00997379e-02  1.00965894e+00 -1.26323111e-01 -1.53294851e+00]
 [ 1.10292917e-01  1.20605640e+00 -1.56982082e-01 -1.86223761e+00]
 [ 1.34414045e-01  1.01296538e+00 -1.94226834e-01 -1.62212597e+00]
 [ 1.54673352e-01  8.20588296e-01 -2.26669353e-01 -1.39573053e+00]], shape=(22, 4), dtype=float64)

batchQVals是

tf.Tensor(
[6.4799747 6.447029  6.40827   6.3626704 6.309024  6.2459106 6.1716595
  6.0843053 5.9815354 5.86063   5.718388  5.551045  5.354171  5.122554
  4.8500633 4.529486  4.1523366 3.7086313 3.186625  2.5725    1.85
  1.       ], shape=(22,), dtype=float32)