r - Keras (R)：训练后移除层

Question

假设我已经用 R 训练了我的 Keras 模型；例如说：

     model<-keras_model_sequential()

     model %>% layer_dense(units=5,activation = "relu",input_shape = c(4))%>% 

     layer_dense(units=Height,activation = "relu",input_shape = c(4)) %>%

     model %>% layer_dense(units=1)

训练完成后，是否可以删除最后一层，同时保持其他所有内容不变？

标签： rtensorflowkerasdeep-learning

Answer 1

如果您先单独定义图层，则可以使用相同的图层创建新模型：

model <- keras_model_sequential()

firstLayer <- layer_dense(units=5,activation = "relu",input_shape = c(4))
secondLayer <- layer_dense(units=Height,activation = "relu",input_shape = c(4))

model %>% firstLayer %>% 
secondLayer %>%
layer_dense(units=1)

新型号：

model2 <- keras_model_sequential()
model2 %>% firstLayer %>%
secondLayer