Trainer¶

Overview¶

The txeo::Trainer class is an abstract base class that provides the interface for training machine learning models in txeo. It handles training/evaluation data, common training parameters, and the training lifecycle.

Derived classes must implement the predict() and train() methods.

Features¶

Abstract base class with pure virtual methods.
Manages training and evaluation datasets.
Supports early stopping.
Tracks whether the model has been trained.

Template Parameter¶

T: The numeric type used in tensors (e.g., float, double).

Constructors¶

Trainer(x_train, y_train, x_eval, y_eval)¶

Initializes the trainer with a data table object.

Trainer(const txeo::DataTable<T> &data);

Public Methods¶

fit(epochs, metric)¶

Trains the model for a fixed number of epochs.

void fit(size_t epochs, txeo::LossFunc metric);

fit(epochs, metric, patience)¶

Trains the model with early stopping.

void fit(size_t epochs, txeo::LossFunc metric, size_t patience);

fit(epochs, metric, patience, type)¶

Trains the model with early stopping and feature normalization.

void fit(size_t epochs, txeo::LossFunc metric, size_t patience, txeo::NormalizationType type);

predict(input)¶

Pure virtual method to generate predictions from a trained model. Must be implemented in derived classes.

txeo::Tensor<T> predict(const txeo::Tensor<T>& input) = 0;

compute_test_loss(txeo::LossFunc metric) const¶

Computes the loss of the trained model for test data.

virtual T compute_test_loss(txeo::LossFunc metric) const;

is_trained()¶

Returns true if the model has been trained.

bool is_trained() const;

data_table()¶

Returns std::DataTable object owned by the trainer.

const txeo::DataTable<T> &data_table() const;

enable_feature_norm()¶

Enables normalization of feature data (input).

void enable_feature_norm(txeo::NormalizationType type);

disable_feature_norm()¶

Disable normalization of feature data (input).

void disable_feature_norm();

Exceptions¶

TrainerError¶

Exception type thrown by Trainer operations.

class TrainerError : public std::runtime_error;

Example Usage¶

class MyTrainer : public txeo::Trainer<float> {
  public:
    using txeo::Trainer<float>::Trainer;

    txeo::Tensor<float> predict(const txeo::Tensor<float>& input) override {
        // your prediction logic
    }

  protected:
    void train(size_t epochs, txeo::LossFunc loss_func) override {
        // your training logic
    }
};

For detailed API references, see individual method documentation at txeo::Predictor.