Module optimizers

Neural network optimizers

This module provides various optimization algorithms for neural networks such as SGD, Adam, RMSProp, Adagrad, etc.

The optimizers in this module are wrappers around the implementations in the scirs2-optim crate, adapted to work with the neural network API.

Re-exports

pub use adagrad::Adagrad;

pub use adam::Adam;

pub use adamw::AdamW;

pub use lr_scheduler_wrapper::with_cosine_annealing;

pub use lr_scheduler_wrapper::with_step_decay;

pub use lr_scheduler_wrapper::LRSchedulerOptimizer;

pub use momentum::MomentumOptimizer;

pub use radam::RAdam;

pub use rmsprop::RMSprop;

pub use sgd::SGD;

Modules

adagrad

Adagrad optimizer implementation

adam

Adam optimizer implementation for neural networks

adamw

AdamW optimizer implementation for neural networks

lr_scheduler_wrapper

Learning rate scheduler integration for optimizers

momentum

Momentum optimizer for neural networks

radam

RAdam (Rectified Adam) optimizer implementation for neural networks

rmsprop

RMSprop optimizer implementation

sgd

Stochastic Gradient Descent optimizer for neural networks

Traits

Optimizer

Trait for neural network optimizers

OptimizerStep

Extension trait for optimizers that can work with model layers