Question

我正在尝试制作多层Keras RNN模型，如下所示：

ttrainx=as.matrix(data.frame(c1=c(0,2,3,4,5,1,2,3,4,5),c2=c(1,3,4,5,6,1,2,3,4,5),c3=c(0,4,5,6,7,1,2,3,4,5)))
ttrainy=as.matrix(data.frame(y1=c(5,6,7,8,9,1,2,3,4,5),y2=c(6,7,8,9,1,2,3,4,5,5),y3=c(0,1,0,0,0,0,0,0,1,1)))

# Initialize model

#========
model <- keras_model_sequential()


model %>%

  layer_masking(mask_value = -1,input_shape = list(NULL,3))
  layer_lstm(units = 10,return_sequences = TRUE) %>% 
  layer_lstm(units = 10,return_sequences = TRUE) %>% 
  layer_lstm(units = 10,return_sequences = TRUE) %>% 
  layer_lstm(units = 10) %>% 
  layer_dropout(rate = 0.5) %>% 
  layer_dense(units =3, activation = 'linear')


# Try using different optimizers and different optimizer configs
model %>% compile(
  loss = 'mse',
  optimizer = 'rmsprop',
  metrics = c('accuracy')
)

# Train model over four epochs
cat('Train...\n')
model %>% fit(
  #x_train, y_train,
  ttrainx,ttrainy,
  batch_size = 5,
  epochs = 30,
  #validation_data = list(x_test, y_test)
  validation_split = 0.2
)

但是我在模型拟合时遇到了以下异常：

py_call_impl（可调用，dots $ args，dots $ keywords）错误： ValueError：检查输入时出错：预期masking_1_input具有3维，但数组的形状为（10，3）

任何人都可以指出为什么我会遇到例外吗？

Keras R API在第一层中制作了一个遮罩模型

0 个答案: