您不能有大小可变的常数。常数始终具有相同的值。您可以做的就是获取
(1,50)常量，然后使用将该常量平铺在TensorFlow中
K.tile。也可以
np.arange代替更好地使用
np.array(list(range(50))。就像是：
from keras.layers.core import Lambdaimport keras.backend as Kdef operateWithConstant(input_batch): tf_constant = K.constant(np.arange(50).reshape((1, 50))) batch_size = K.shape(input_batch)[0] tiled_constant = K.tile(tf_constant, (batch_size, 1)) # Do some operation with tiled_constant and input_batch result = ... return resultinput_batch = Input(...)input_operated = Lambda(operateWithConstant)(input_batch)# continue...
