【Tensorflow 2.7.0 Error】

Tensorflow 2.7.0 Error

1. TypeError: cannot unpack non-iterable function object

# fashion_mnist.load_data为函数对象，需加()调用才能返回值
(x_train_all,y_train_all),(x_test,y_test) = fashion_mnist.load_data
# 代码修改如下
(x_train_all,y_train_all),(x_test,y_test) = fashion_mnist.load_data()

2. The truth value of a Series is ambiguous. Use a.empty, a.bool(), a.item(), a.any() or a.all().

# 将feature和label合并为Dataset对象，需要用(feature,label)打包
    dataset = tf.data.Dataset.from_tensor_slices(dict(data_df), label_df)
# 代码修改如下,其中dict（data_df)为feature
tf.data.Dataset.from_tensor_slices((dict(data_df), label_df))

3. ValueError: Dimensions must be equal, but are 64 and 2 for ‘{{node mean_squared_error/SquaredDifference}} = SquaredDifference[T=DT_FLOAT](functional_41/complex_conv_266/activation_266/SelectV2_1, mean_squared_error/Cast)’ with input shapes: [?,28,28,64], [?,28,28,2]

错误分析： model.fit时报错mean_squared_error节点的输入输出形状不一致，打开model的定义发现是自己输出的最后一层的filters设置错误，train_images的形状为（28，28，2），因此输出的最后一层形状也应为（28，28，2），设置filters=2。