Struct CacheFriendlyAdam 

pub struct CacheFriendlyAdam { /* private fields */ }

Cache-friendly Adam optimizer implementation.

This optimizer uses blocked processing and optimized memory layouts to minimize cache misses and improve performance.

Implementations

impl CacheFriendlyAdam

pub fn new(lr: f32, betas: (f32, f32), eps: f32, weight_decay: f32) -> Self

Creates a new cache-friendly Adam optimizer.

pub fn with_cache_config( lr: f32, betas: (f32, f32), eps: f32, weight_decay: f32, cache_config: CacheConfig, ) -> Self

Creates a cache-friendly Adam optimizer with custom cache configuration.

pub fn cache_stats(&self) -> CacheStats

Gets cache utilization statistics.

pub fn cleanup_unused_params(&mut self, steps_threshold: usize)

Performs garbage collection on unused parameters.

Trait Implementations

impl Debug for CacheFriendlyAdam

fn fmt(&self, f: &mut Formatter<'_>) -> Result

Formats the value using the given formatter.

impl Optimizer for CacheFriendlyAdam

fn update(&mut self, parameter: &mut Tensor, grad: &Tensor) -> Result<()>

Updates a parameter based on its gradient.

fn zero_grad(&mut self)

Clears all accumulated gradients.

fn step(&mut self)

Performs a single optimization step.

fn get_lr(&self) -> f32

Gets the current learning rate.

fn set_lr(&mut self, lr: f32)

Sets a new learning rate.

fn accumulate_grad( &mut self, parameter: &mut Tensor, grad: &Tensor, ) -> Result<(), TrustformersError>

Accumulates gradients for gradient accumulation.

fn apply_accumulated_grads( &mut self, accumulation_steps: usize, ) -> Result<(), TrustformersError>

Applies accumulated gradients after gradient accumulation.