Can't build model on Rstudio

[edwardfung2018]

Error in py_call_impl(callable, dots$args, dots$keywords) :
ValueError: Layer vgg16 was called with an input that isn't a symbolic tensor. Received type: <class 'keras.engine.sequential.Sequential'>. Full input: [<keras.engine.sequential.Sequential object at 0x000000002DFAB9E8>]. All inputs to the layer should be tensors.

[skeydan]

Hi, please post a minimal reproducible example.

[edwardfung2018]

I follow this example https://tensorflow.rstudio.com/keras/ . When I run following code the error was shown.

model <- keras_model_sequential()
model %>%
layer_dense(units = 256, activation = 'relu', input_shape = c(784)) %>%
layer_dropout(rate = 0.4) %>%
layer_dense(units = 128, activation = 'relu') %>%
layer_dropout(rate = 0.3) %>%
layer_dense(units = 10, activation = 'softmax')

[skeydan]

Please post the complete code you're executing, and point out at which point the error occurs.

[edwardfung2018]

install.packages("keras")
library(keras)
install_keras()

library(keras)
mnist <- dataset_mnist()
x_train <- mnist$train$x
y_train <- mnist$train$y
x_test <- mnist$test$x
y_test <- mnist$test$y

x_train <- array_reshape(x_train, c(nrow(x_train), 784))
x_test <- array_reshape(x_test, c(nrow(x_test), 784))

x_train <- x_train / 255
x_test <- x_test / 255

y_train <- to_categorical(y_train, 10)
y_test <- to_categorical(y_test, 10)

model <- keras_model_sequential()
model %>%
layer_dense(units = 256, activation = 'relu', input_shape = c(784)) %>%
layer_dropout(rate = 0.4) %>%
layer_dense(units = 128, activation = 'relu') %>%
layer_dropout(rate = 0.3) %>%
layer_dense(units = 10, activation = 'softmax')

All script is as above. when it run to the last part "model" then error was shown.

[skeydan]

Hi,

that code alone can't be throwing the above error. Try restarting R and then, just start with library(keras) and what follows.

The reason I'm saying this is that the error says something about "Layer vgg16 ".
VGG16 is a pretrained model that can be obtained like so

model <- application_vgg16()

which does not appear in the above code.

[cak925]

I am also getting an error similar to this when I run:

library(keras)

model <- keras_model_sequential()

model %>%
layer_flatten( input_shape = c(28, 28)) %>%
layer_dense(units = 128, activation = 'relu') %>%
layer_dense(units = 10, activation = 'softmax')

I obtain the error:
" Error in py_call_impl(callable, dots$args, dots$keywords) :
ValueError: Layer flatten_1 was called with an input that isn't a symbolic tensor. Received type: <class 'keras.engine.sequential.Sequential'>. Full input: [<keras.engine.sequential.Sequential object at 0x0000000006BFA978>]. All inputs to the layer should be tensors. "

[cak925]

I have also tried to run the same example as edwardfung2018, and received the exact error he did.

[msmith01]

Obtaining the following error after running:

library(keras)
max_features <- 10000
maxlen <- 20

imdb <- dataset_imdb(num_words = max_features)

c(c(x_train, y_train), c(x_test, y_test)) %<-% imdb
x_train <- pad_sequences(x_train, maxlen = maxlen)
x_test <- pad_sequences(x_test, maxlen = maxlen)

model <- keras_model_sequential() %>%
  layer_embedding(input_dim = 10000, output_dim = 8,
                  input_length = maxlen) %>%
  layer_flatten() %>%
  layer_dense(units = 1, activation = "sigmoid")

Error:

Error in py_call_impl(callable, dots$args, dots$keywords) : 
 ValueError: Layer embedding_1 was called with an input that isn't a symbolic tensor. Received type: <class 'keras.engine.sequential.Sequential'>. Full input: [<keras.engine.sequential.Sequential object at 0x000000001854BB70>]. All inputs to the layer should be tensors. 

My example comes from the book "Deep Learning with R" - Page 176; Part " Listing 6.7 Using an embedding layer and classifier on the IMDB data "

[msmith01]

What appeared to work for me is to reinstall keras. I think the reticulate package is an older version which might be causing problems. Running the following solved the issue for me

require(devtools)
install_github("rstudio/reticulate")
install_github("rstudio/tensorflow")
install_github("rstudio/keras")

library(keras)
install_keras()

[Divya-Swaroopini]

I've got the exact same error now and none of the solutions seem to be working for me

[t-kalinowski]

Hi, this thread is quite old, and likely stale. Can you please open a new issue?