Module optimizer 

Re-exports

pub use adagrad::AdaGrad;

pub use adagrad::AdaGradConfig;

pub use adamw::AdamW;

pub use adamw::AdamWConfig;

pub use grad_accumulator::GradAccumulator;

pub use grad_clip::clip_grad_norm;

pub use grad_clip::clip_grad_norm_per_param;

pub use grad_clip::clip_grad_value;

pub use grad_scaler::GradScaler;

pub use lamb::Lamb;

pub use lamb::LambConfig;

pub use lr_schedule::DecayShape;

pub use lr_schedule::LrSchedule;

pub use sgd::Sgd;

pub use sgd::SgdConfig;

pub use traits::Optimizer;

Modules

adagrad

AdaGrad optimizer

adamw

AdamW optimizer

grad_accumulator

Gradient accumulation across micro-batches

grad_clip

Gradient clipping utilities

grad_scaler

Dynamic loss scaling for FP16 mixed precision training

lamb

LAMB optimizer (Layer-wise Adaptive Moments for Batch training)

lr_schedule

Learning rate schedulers

sgd

SGD optimizer with momentum

traits

Optimizer trait abstraction