How can I find the number of epochs for which keras model was trained?

1. I use callback_early_stopping() to stop the training early to avoid overfitting.

2. I have been using callback_csv_logger() to log training performances. But sometimes, I train 100s of keras models and it does not make sense to log entire training just to know the number of epochs for which each model.

library(keras)
library(kerasR)
library(tidyverse)


# Data
x = matrix(data = runif(30000), nrow = 10000, ncol = 3)
y = ifelse(rowSums(x) > 1.5   runif(10000), 1, 0)
y = to_categorical(y)

# keras model
model <- keras_model_sequential() %>%   
  layer_dense(units = 50, activation = "relu", input_shape = ncol(x)) %>%
  layer_dense(units = ncol(y), activation = "softmax")

model %>%
  compile(loss = "categorical_crossentropy", 
          optimizer = optimizer_rmsprop(), 
          metrics = "accuracy")

model %>% 
  fit(x, y, 
      epochs = 1000,
      batch_size = 128,
      validation_split = 0.2, 
      callbacks = callback_early_stopping(monitor = "val_loss", patience = 5),
      verbose = 1)

CodePudding user response：

I use tensorflow keras in python however, my initial search would be in the history that holds all information from the relevant logging from the model after fitting (loss, validation loss, accuracy, F1 etc)

I suspect this is the same in R -

as per: https://keras.rstudio.com/articles/training_visualization.html

simply assign a history variable to your model fit call like:

history <- model %>% 
  fit(x, y, 
      epochs = 1000,
      batch_size = 128,
      validation_split = 0.2, 
      callbacks = callback_early_stopping(monitor = "val_loss", patience = 5),
      verbose = 1)

convert history to dataframe (as.data.frame(history)) within which you will find your metrics - the length of the metrics are the same as the number of epochs the model was trained on

CodePudding user response：

To print the number of epochs (whetevere you want) you can use a callback. Here's an example:

class print_log_Callback(Callback):
  def __init__(self, logpath, steps):
    self.logpath = logpath
    self.losslst = np.zeros(steps)

  def on_train_batch_end(self, batch, logs=None):
    self.losslst[batch] = logs["loss"]
    with open(logpath, 'a') as writefile:
      with redirect_stdout(writefile):
        print("For batch {}, loss is {:7.2f}.".format(batch, logs["loss"]))
        writefile.write("\n")

  def on_test_batch_end(self, batch, logs=None):
    with open(logpath, 'a') as writefile:
      with redirect_stdout(writefile):
        print("For batch {}, val_loss is {:7.2f}.".format(batch, logs["loss"]))
        writefile.write("\n")

  def on_epoch_end(self, epoch, logs=None):
    with open(logpath, 'a') as writefile:
      with redirect_stdout(writefile):
        print("The val_loss  for epoch {} is {:7.2f}.".format(epoch, logs['val_loss']))
        writefile.write("\n")
        print("The mean train loss is: ", np.mean(self.losslst))
        writefile.write("\n")
        writefile.write("\n")

    self.losslst = np.zeros(steps)

you call it like this:

print_log_Callback(logpath=logpath, steps=int(steps))

where logpath is the path of the text file where you are writing the code and steps is the number of steps.

This callback basically prints on a text file the entire history of the network.

loss after every batch and after every epoch end.

If you need only the epoch you could use just the method on_epoch_end and remove everything else.

If you want to print the loss after every epoch you can use this modified version:

class print_log_Callback(Callback):
  def __init__(self, logpath, steps):
    self.logpath = logpath
    self.losslst = np.zeros(steps)

  def on_train_batch_end(self, batch, logs=None):
    self.losslst[batch] = logs["loss"]

  def on_epoch_end(self, epoch, logs=None):
    with open(logpath, 'a') as writefile:
      with redirect_stdout(writefile):
        print("The val_loss  for epoch {} is {:7.2f}.".format(epoch, logs['val_loss']))
        writefile.write("\n")
        print("The mean train loss is: ", np.mean(self.losslst))
        writefile.write("\n")
        writefile.write("\n")

    self.losslst = np.zeros(steps)

you can modify this callback to print also the metric: just print logs["accuracy"] for example.