plot_clr_history: Simple plotting utility

In lorenzwalthert/KerasMisc: Add-ons for Keras

Description

Simple plotting utility

Usage

plot_clr_history(
  callback_clr,
  granularity = "epoch",
  backend = "ggplot2",
  trans_y_axis = "identity"
)

Arguments

callback_clr	An object of class CyclicLR.
granularity	Either "epoch" or "iteration". We advise to use epoch as we find it easier to work with. The plot will look very similar (except for the x-axis scaling) for both options as long as you choosed step_size in new_callback_cyclical_learning_rate() to be more iterations than one epoch has.
backend	Either "base" for base R or "ggplot2".
trans_y_axis	Value passed to ggplot2::scale_y_continuous() as the trans argument. Only supported for backend = "ggplot2".

Examples

library(keras)
dataset <- dataset_boston_housing()
c(c(train_data, train_targets), c(test_data, test_targets)) %<-% dataset

mean <- apply(train_data, 2, mean)
std <- apply(train_data, 2, sd)
train_data <- scale(train_data, center = mean, scale = std)
test_data <- scale(test_data, center = mean, scale = std)


model <- keras_model_sequential() %>%
  layer_dense(
    units = 64, activation = "relu",
    input_shape = dim(train_data)[[2]]
  ) %>%
  layer_dense(units = 64, activation = "relu") %>%
  layer_dense(units = 1)
model %>% compile(
  optimizer = optimizer_rmsprop(lr = 0.001),
  loss = "mse",
  metrics = c("mae")
)

callback_clr <- new_callback_cyclical_learning_rate(
  step_size = 32,
  base_lr = 0.001,
  max_lr = 0.006,
  gamma = 0.99,
  mode = "exp_range"
)
model %>% fit(
  train_data, train_targets,
  validation_data = list(test_data, test_targets),
  epochs = 10, verbose = 1,
  callbacks = list(callback_clr)
)
callback_clr$history
plot_clr_history(callback_clr, backend = "base")

lorenzwalthert/KerasMisc documentation built on May 7, 2021, 6:31 a.m.