R Keras: Error in basic tutorial regression. Error in py_call_impl(callable, dots$args, dots$keywords) : ValueError: in user code:


I’ve been following a couple of tensorflow tutorials that don’t work.


Here is some sample code that is very similar to what’s in the tutorials.

x <-rnorm(1000) #input variables
y <- rnorm(1000)    #input variables
z <-x+y+rnorm(1000) #output variable
df <-data.frame(x=x,y=y,z=z)

model <- keras_model_sequential() %>%
layer_dense(units = 8,activation = "relu",input_shape = 2) %>%
layer_dense(units = 8,activation = "relu") %>%
layer_dense(units = 1,activation = "relu") 

model %>% compile(
loss = "mse",
  optimizer = optimizer_adam(),
metrics = list("mean_absolute_error"))

model %>% fit(df[,1:2],df[,3], epochs = 20)

When I run it, I get this error:

Error in py_call_impl(callable, dots$args, dots$keywords) : 
  ValueError: in user code:

    C:\Users\User\AppData\Local\R-MINI~1\envs\R-RETI~1\lib\site-packages\keras\engine\training.py:853 train_function  *
        return step_function(self, iterator)
    C:\Users\User\AppData\Local\R-MINI~1\envs\R-RETI~1\lib\site-packages\keras\engine\training.py:842 step_function  **
        outputs = model.distribute_strategy.run(run_step, args=(data,))
    C:\Users\User\AppData\Local\R-MINI~1\envs\R-RETI~1\lib\site-packages\tensorflow\python\distribute\distribute_lib.py:1286 run
        return self._extended.call_for_each_replica(fn, args=args, kwargs=kwargs)
    C:\Users\User\AppData\Local\R-MINI~1\envs\R-RETI~1\lib\site-packages\tensorflow\python\distribute\distribute_lib.py:2849 call_for_each_replica
        return self._call_for_each_replica(fn, args, kwargs)
    C:\Users\User\AppData\Local\R-MINI~1\envs\R-RETI~1\lib\site-packages\tensorflow\python\distribute\distribute_lib.py:3632 _call_for_each_replica
        return fn(*args, **kwargs)

I am using tensorflow 2.6

What in this code could be causing me this error, and how would I fix it?


You probably want to pass an array to the fit() method for the input training data, right now you are passing in a data.frame. Converting df to a matrix/R array in your example makes everything work:

x <- rnorm(1000) #input variables
y <- rnorm(1000)    #input variables
z <- x + y + rnorm(1000) #output variable
mat <- cbind(x, y, z)

model <- keras_model_sequential(input_shape = 2) %>%
  layer_dense(8, activation = "relu") %>%
  layer_dense(8, activation = "relu") %>%
  layer_dense(1, activation = "relu")
#> Loaded Tensorflow version 2.6.0

model %>% compile(
  loss = "mse",
  optimizer = optimizer_adam(),
  metrics = list("mean_absolute_error")

model %>% fit(mat[, 1:2], mat[, 3])

Created on 2021-10-05 by the reprex package (v2.0.1)

Answered By – t-kalinowski

This Answer collected from stackoverflow, is licensed under cc by-sa 2.5 , cc by-sa 3.0 and cc by-sa 4.0

Leave a Reply

(*) Required, Your email will not be published