fit.reservr_keras_model {reservr}

Fit a neural network based distribution model to data

Description

This function delegates most work to keras3::fit.keras.src.models.model.Model() and performs additional consistency checks to make sure tf_compile_model() was called with the appropriate options to support fitting the observations y as well as automatically converting y to a n x 6 matrix needed by the compiled loss function.

Usage

## S3 method for class 'reservr_keras_model'
fit(
  object,
  x,
  y,
  batch_size = NULL,
  epochs = 10,
  verbose = getOption("keras.fit_verbose", default = 1),
  callbacks = NULL,
  view_metrics = getOption("keras.view_metrics", default = "auto"),
  validation_split = 0,
  validation_data = NULL,
  shuffle = TRUE,
  class_weight = NULL,
  sample_weight = NULL,
  initial_epoch = 0,
  steps_per_epoch = NULL,
  validation_steps = NULL,
  ...
)

Arguments

object

A compiled reservr_keras_model as obtained by tf_compile_model().

x

A list of input tensors (predictors)

y

A trunc_obs tibble of observed outcomes, or something convertible via as_trunc_obs().

batch_size

Integer or NULL. Number of samples per gradient update. If unspecified, batch_size will default to 32. Do not specify the batch_size if your data is in the form of TF Datasets or generators, (since they generate batches).

epochs

Integer. Number of epochs to train the model. An epoch is an iteration over the entire x and y data provided (unless the steps_per_epoch flag is set to something other than NULL). Note that in conjunction with initial_epoch, epochs is to be understood as "final epoch". The model is not trained for a number of iterations given by epochs, but merely until the epoch of index epochs is reached.

verbose

"auto", 0, 1, or 2. Verbosity mode. 0 = silent, 1 = progress bar, 2 = one line per epoch. "auto" becomes 1 for most cases, 2 if in a knitr render or running on a distributed training server. Note that the progress bar is not particularly useful when logged to a file, so verbose=2 is recommended when not running interactively (e.g., in a production environment). Defaults to "auto".

callbacks

List of Callback() instances. List of callbacks to apply during training. See ⁠callback_*⁠.

view_metrics

View realtime plot of training metrics (by epoch). The default ("auto") will display the plot when running within RStudio, metrics were specified during model compile(), epochs > 1 and verbose > 0. Set the global options(keras.view_metrics = ) option to establish a different default.

validation_split

Float between 0 and 1. Fraction of the training data to be used as validation data. The model will set apart this fraction of the training data, will not train on it, and will evaluate the loss and any model metrics on this data at the end of each epoch. The validation data is selected from the last samples in the x and y data provided, before shuffling. This argument is not supported when x is a TF Dataset or generator. If both validation_data and validation_split are provided, validation_data will override validation_split.

validation_data

Data on which to evaluate the loss and any model metrics at the end of each epoch. The model will not be trained on this data. Thus, note the fact that the validation loss of data provided using validation_split or validation_data is not affected by regularization layers like noise and dropout. validation_data will override validation_split. It could be:

• A tuple ⁠(x_val, y_val)⁠ of arrays or tensors.

• A tuple ⁠(x_val, y_val, val_sample_weights)⁠ of arrays.

• A generator returning ⁠(inputs, targets)⁠ or ⁠(inputs, targets, sample_weights)⁠.

shuffle

Boolean, whether to shuffle the training data before each epoch. This argument is ignored when x is a generator or a TF Dataset.

class_weight

Optional named list mapping class indices (integers, 0-based) to a weight (float) value, used for weighting the loss function (during training only). This can be useful to tell the model to "pay more attention" to samples from an under-represented class. When class_weight is specified and targets have a rank of 2 or greater, either y must be one-hot encoded, or an explicit final dimension of 1 must be included for sparse class labels.

sample_weight

Optional array of weights for the training samples, used for weighting the loss function (during training only). You can either pass a flat (1D) array/vector with the same length as the input samples (1:1 mapping between weights and samples), or in the case of temporal data, you can pass a 2D array (matrix) with shape ⁠(samples, sequence_length)⁠, to apply a different weight to every timestep of every sample. This argument is not supported when x is a TF Dataset or generator, instead provide the sample_weights as the third element of x. Note that sample weighting does not apply to metrics specified via the metrics argument in compile(). To apply sample weighting to your metrics, you can specify them via the weighted_metrics in compile() instead.

initial_epoch

Integer. Epoch at which to start training (useful for resuming a previous training run).

steps_per_epoch

Integer or NULL. Total number of steps (batches of samples) before declaring one epoch finished and starting the next epoch. When training with input tensors such as backend-native tensors, the default NULL is equal to the number of samples in your dataset divided by the batch size, or 1 if that cannot be determined. If x is a TF Dataset, and steps_per_epoch is NULL, the epoch will run until the input dataset is exhausted. When passing an infinitely repeating dataset, you must specify the steps_per_epoch argument. If steps_per_epoch = -1 the training will run indefinitely with an infinitely repeating dataset.

validation_steps

Only relevant if validation_data is provided. Total number of steps (batches of samples) to draw before stopping when performing validation at the end of every epoch. If validation_steps is NULL, validation will run until the validation_data dataset is exhausted. In the case of an infinitely repeated dataset, it will run into an infinite loop. If validation_steps is specified and only part of the dataset will be consumed, the evaluation will start from the beginning of the dataset at each epoch. This ensures that the same validation samples are used every time.

...

Unused. If old arguments are supplied, an error message will be raised informing how to fix the issue.

Details

Additionally, the default batch_size is min(nrow(y), 10000) instead of keras default of 32 because the latter is a very bad choice for fitting most distributions since the involved loss is much less stable than typical losses used in machine learning, leading to divergence for small batch sizes.

Value

A history object that contains all information collected during training. The model object will be updated in-place as a side-effect.

See Also

predict.reservr_keras_model tf_compile_model keras3::fit.keras.src.models.model.Model

Examples

dist <- dist_exponential()
params <- list(rate = 1.0)
N <- 100L
rand_input <- runif(N)
x <- dist$sample(N, with_params = params)

if (interactive()) {
  tf_in <- keras3::layer_input(1L)
  mod <- tf_compile_model(
    inputs = list(tf_in),
    intermediate_output = tf_in,
    dist = dist,
    optimizer = keras3::optimizer_adam(),
    censoring = FALSE,
    truncation = FALSE
  )

  tf_fit <- fit(
    object = mod,
    x = k_matrix(rand_input),
    y = x,
    epochs = 10L,
    callbacks = list(
      callback_debug_dist_gradients(mod, k_matrix(rand_input), x, keep_grads = TRUE)
    )
  )
}