adam = keras.optimizers.Adam(lr=0.0001, beta_1=0.5, beta_2=0.9, epsilon=1.0e-08, clipnorm=50)
model.compile(optimizer=adam, loss='mse')
`
MXNetError: [21:55:12] src/operator/nn/deconvolution.cc:379: Check failed: param.kernel.ndim() == param.dilate.ndim() (3 vs. 2) : Dilate must have the same number of dimensions with kernel_size,but kernel_size is set to [3,4,4] while dilate is [1,1]
Conv3DTranspose seems to have a bug
` model = Sequential()
model.add(Conv3D(filters=128, kernel_size=(4,4,4), strides=(2,2,2), padding='same', input_shape=(10, 64, 64, 1)))
model.add(Conv3DTranspose(filters=1, kernel_size=(3,4,4), strides=(1,2,2), padding='same'))
adam = keras.optimizers.Adam(lr=0.0001, beta_1=0.5, beta_2=0.9, epsilon=1.0e-08, clipnorm=50) model.compile(optimizer=adam, loss='mse') `