tests/testthat/test-load_save.R

In kerasR: R Interface to the Keras Deep Learning Library

library(kerasR)

context("Testing loading and saving data")

check_keras_available <- function() {
  if (!keras_available(silent = TRUE)) {
    skip("Keras is not available on this system.")
  }
}

test_that("loading and saving", {
  skip_on_cran()
  check_keras_available()

  # X_train <- matrix(rnorm(100 * 10), nrow = 100)
  # Y_train <- to_categorical(matrix(sample(0:2, 100, TRUE), ncol = 1), 3)
  mod <- Sequential()
  mod$add(Dense(units = 50, input_shape = 10))
  mod$add(Dropout(rate = 0.5))
  mod$add(Activation("relu"))
  mod$add(Dense(units = 3))
  mod$add(ActivityRegularization(l1 = 1))
  mod$add(Activation("softmax"))
  keras_compile(mod,  loss = 'categorical_crossentropy', optimizer = RMSprop())
  # keras_fit(mod, X_train, Y_train, batch_size = 32, epochs = 5,
  #           verbose = 0, validation_split = 0.2)

  testthat::expect_false(mod$stateful)

  # save/load the entire model object
  keras_save(mod, tf <- tempfile())
  mod2 <- keras_load(tf)

  # save/load just the weights file
  keras_save_weights(mod, tf <- tempfile())
  keras_load_weights(mod, tf)

  # save/load just the architecture (as human readable json)
  tf <- tempfile(fileext = ".json")
  keras_model_to_json(mod, tf)
  # cat(readLines(tf))
  mod3 <- keras_model_from_json(tf)
})