turgut090
commented
4 years ago Hi, I think this is due to keras version. Because I tried to install tf-nightly and keras from R side and as a result the Keras version was 2.3.0-tf and tf version was 2.2-nightly. But if you import packages via reticulate the following will work fine:
keras = reticulate::import('keras')
tensorflow = reticulate::import('tensorflow')
> keras = reticulate::import('keras')
> keras$`__version__`
[1] "2.3.1"
> tensorflow = reticulate::import('tensorflow')
> tensorflow$`__version__`
[1] "2.2.0-dev20200329"Later, this will work fine but this is pythonic:
model = keras$Sequential()
model$add(keras$layers$Dense(units=1L))
model$compile(loss=keras$losses$binary_crossentropy,
optimizer=keras$optimizers$SGD(lr=0.01, momentum=0.9, nesterov=T))
model$fit(x_data, y_data, epochs=5L, batch_size=32L)
skeydan
djduncan585
I'm attempting to run the text classification tutorial located at https://tensorflow.rstudio.com/tutorials/beginners/basic-ml/tutorial_basic_text_classification/ inside of R Studio. Version of TensorFlow for R Studio installed is 2.2.
I attempt to run the section of the code that says history <- model %>% fit( training$text, as.numeric(training$tag == "pos"), epochs = 10, batch_size = 512, validation_split = 0.2, verbose=2 ) and get back the error Error in py_call_impl(callable, dots$args, dots$keywords) : RuntimeError: Evaluation error: invalid argument type.
I've reinstalled Keras and updated TensorFlow. I've reinstalled Reticulate. Any ideas?