TensorFlow：tensorflow.keras.Model.fit（）报错： TypeError: Failed to convert elements of xxx

一、问题

在使用TensorFlow进行数据训练的时候，报了下面这样的一个错误。

代码如下：

import tensorflow as tf
from utilz import *
import numpy as np

acoustic = load_features('C:/Test/MSA Datasets/data/acoustic_wav2vec.pkl')
label = load_features('C:/Test/MSA Datasets/data/labels.pkl')

shapeReturn = np.shape(acoustic['train'])
print(shapeReturn)

x = tf.keras.layers.Input((128, 512))
h = tf.keras.layers.LSTM(32, dropout=0.5)(x)
res = tf.keras.layers.Dense(3, 'softmax')(h)

model = tf.keras.Model(inputs=x, outputs=res)
# 问题出现在下面这一行
model.compile(optimizer='Adam', loss=tf.keras.losses.SparseCategoricalCrossentropy)
model.summary()
print(model.summary())
output_layer = model.layers[-1]
print(output_layer.activation)



callback_list = [tf.keras.callbacks.ModelCheckpoint(filepath='C:/Test/MSA Datasets/res_tmp/model.tf', monitor='val_