Struct GradientMonitor 

pub struct GradientMonitor {
    pub gradient_norms: Vec<f64>,
    pub gradient_means: Vec<f64>,
    pub gradient_stds: Vec<f64>,
    pub vanishing_count: usize,
    pub exploding_count: usize,
    /* private fields */
}

Gradient flow monitor for tracking gradient statistics during training.

This callback tracks gradient norms, mean, std, and identifies vanishing/exploding gradients. Useful for debugging training issues and understanding gradient flow through the network.

Example

use tensorlogic_train::{GradientMonitor, CallbackList};

let mut callbacks = CallbackList::new();
callbacks.add(Box::new(GradientMonitor::new(
    10,      // log_frequency
    1e-7,    // vanishing_threshold
    100.0,   // exploding_threshold
)));

Fields

gradient_norms: Vec<f64>

History of gradient norms.

gradient_means: Vec<f64>

History of gradient means.

gradient_stds: Vec<f64>

History of gradient stds.

vanishing_count: usize

Count of vanishing gradient warnings.

exploding_count: usize

Count of exploding gradient warnings.

Implementations

impl GradientMonitor

pub fn new( log_frequency: usize, vanishing_threshold: f64, exploding_threshold: f64, ) -> Self

Create a new gradient monitor.

Arguments

• log_frequency - Log statistics every N batches
• vanishing_threshold - Threshold below which gradients are considered vanishing
• exploding_threshold - Threshold above which gradients are considered exploding

pub fn summary(&self) -> GradientSummary

Get summary statistics.

Trait Implementations

impl Callback for GradientMonitor

fn on_batch_end( &mut self, _batch: usize, state: &TrainingState, ) -> TrainResult<()>

Called at the end of a batch.

fn on_train_end(&mut self, _state: &TrainingState) -> TrainResult<()>

Called at the end of training.

fn on_train_begin(&mut self, _state: &TrainingState) -> TrainResult<()>

Called at the beginning of training.

fn on_epoch_begin( &mut self, _epoch: usize, _state: &TrainingState, ) -> TrainResult<()>

Called at the beginning of an epoch.

fn on_epoch_end( &mut self, _epoch: usize, _state: &TrainingState, ) -> TrainResult<()>

Called at the end of an epoch.

fn on_batch_begin( &mut self, _batch: usize, _state: &TrainingState, ) -> TrainResult<()>

Called at the beginning of a batch.

fn on_validation_end(&mut self, _state: &TrainingState) -> TrainResult<()>

Called after validation.

fn should_stop(&self) -> bool

Check if training should stop early.