张力流梯度凸轮-型号.fit()-值错误:形状(None,1)和(None,2)不兼容

作为作业4,Coursera CV TF course的一部分,我的代码在 model.fit()

model.compile(loss='categorical_crossentropy',metrics=
   ['accuracy'],optimizer=tf.keras.optimizers.RMSprop(lr=0.001))
# shuffle and create batches before training

model.fit(train_batches,epochs=25)

有错误:

ValueError: Shapes (None, 1) and (None, 2) are incompatible

有没有任何迹象表明问题可能来自哪里?我怀疑格式或类型不正确 train_batches :

train_data = tfds.load('cats_vs_dogs', split='train[:80%]', as_supervised=True) 
augmented_training_data = train_data.map(augmentimages)
train_batches = augmented_training_data.batch(32)