Struct LSTMBatchTrainer

pub struct LSTMBatchTrainer<L: LossFunction, O: Optimizer> {
    pub network: LSTMNetwork,
    pub loss_function: L,
    pub optimizer: O,
    pub config: TrainingConfig,
    pub metrics_history: Vec<TrainingMetrics>,
}

Batch trainer for LSTM networks with configurable loss and optimizer Processes multiple sequences simultaneously for improved performance

Fields

network: LSTMNetwork

loss_function: L

optimizer: O

config: TrainingConfig

metrics_history: Vec<TrainingMetrics>

Implementations

impl<L: LossFunction, O: Optimizer> LSTMBatchTrainer<L, O>

pub fn new(network: LSTMNetwork, loss_function: L, optimizer: O) -> Self

pub fn with_config(self, config: TrainingConfig) -> Self

pub fn train_batch( &mut self, batch_inputs: &[Vec<Array2<f64>>], batch_targets: &[Vec<Array2<f64>>], ) -> f64

Train on a batch of sequences using batch processing

Arguments

• batch_inputs - Vector of input sequences, each sequence is Vec<Array2>
• batch_targets - Vector of target sequences, each sequence is Vec<Array2>

Returns

• Average loss across the batch

pub fn train( &mut self, train_data: &[(Vec<Array2<f64>>, Vec<Array2<f64>>)], validation_data: Option<&[(Vec<Array2<f64>>, Vec<Array2<f64>>)]>, batch_size: usize, )

Train for multiple epochs with batch processing

Arguments

• train_data - Vector of (input_sequences, target_sequences) tuples for training
• validation_data - Optional validation data
• batch_size - Number of sequences to process in each batch

pub fn evaluate_batch( &mut self, data: &[(Vec<Array2<f64>>, Vec<Array2<f64>>)], batch_size: usize, ) -> f64

Evaluate model performance using batch processing

pub fn predict_batch( &mut self, inputs: &[Vec<Array2<f64>>], ) -> Vec<Vec<Array2<f64>>>

Generate predictions using batch processing

pub fn get_latest_metrics(&self) -> Option<&TrainingMetrics>

pub fn get_metrics_history(&self) -> &[TrainingMetrics]

pub fn set_training_mode(&mut self, training: bool)