No gradients provided for any variable in R

Question

I am using R for deep learning with the MNIST dataset.

I have written this code to store the training and testing data, and define and fit the model:

library(keras)  

#Obtain data 
mnist <- dataset_mnist()  
train_data <- mnist$train$x 
train_labels <- mnist$train$y  
test_data <- mnist$test$x 
test_labels <- mnist$test$y  

#Reshape & normalize
train_data <- array_reshape(train_data,c(nrow(train_data), 784)) 
train_data <- train_data / 255 
test_data <- array_reshape(test_data,c(nrow(test_data), 784)) 
test_data <- test_data / 255

#One hot encoding train_labels <- to_categorical(train_labels, 10) 
test_labels <- to_categorical(test_labels, 10)

#Model 
model <- keras_model_sequential() 
model %>%   layer_dense(units=128,activation="relu", input_shape=c(784)) %>%          
            layer_dropout(rate=0.3) %>%         
            layer_dense(units=64,activation="relu") %>%          
            layer_dropout(rate=0.2) %>%         
            layer_dense(units=10,activation="softmax") 

#Compile
model %>% compile(loss="categorical_crossentropy",             
                    optimizer="rmsprop",             
                    metrics="accuracy")  
    
#Train
history <- model %>% fit(train_data,                        
                        train_labels,                           
                        epochs=10,                      
                        batch_size=784,                         
                        validation_split=0.2,                       
                        verbose=2)  
    
#Evaluation and prediction 
model %>% evaluate(test_data, test_labels) 
pred <- model %>% predict(test_data) 
print(table(Predicted=pred, Actual=test_labels))

When running it in R studio, the following error occurs:

ValueError: No gradients provided for any variable: (['dense_124/kernel:0', 'dense_124/bias:0', 'dense_123/kernel:0', 'dense_123/bias:0', 'dense_122/kernel:0', 'dense_122/bias:0'],). Provided `grads_and_vars` is ((None, <tf.Variable 'dense_124/kernel:0' shape=(784, 128) dtype=float32>), (None, <tf.Variable 'dense_124/bias:0' shape=(128,) dtype=float32>), (None, <tf.Variable 'dense_123/kernel:0' shape=(128, 64) dtype=float32>), (None, <tf.Variable 'dense_123/bias:0' shape=(64,) dtype=float32>), (None, <tf.Variable 'dense_122/kernel:0' shape=(64, 10) dtype=float32>), (None, <tf.Variable 'dense_122/bias:0' shape=(10,) dtype=float32>)).

I think the problem may be with the conflicting shapes of the input data and the input, but no idea how to solve this.

Thanks for help!