Struct Adam 

pub struct Adam {
    pub learning_rate: LearningRate,
    pub betas: (f64, f64),
    pub eps: f64,
    pub bias_correction: bool,
    pub weight_decay: f64,
    /* private fields */
}

Adam, as MLX’s optimizers.Adam (bias correction off unless enabled, as in MLX), and AdamW (decoupled weight decay) with Adam::weight_decay.

Fields

learning_rate: LearningRate

The learning rate.

betas: (f64, f64)

Decay rates of the first and second moment estimates.

eps: f64

Added to the denominator for numerical stability.

bias_correction: bool

Whether to correct the moment estimates’ bias toward zero.

weight_decay: f64

Decoupled weight decay (AdamW); 0 for Adam.

Implementations

impl Adam

pub fn new(learning_rate: impl Into<LearningRate>) -> Self

Adam with betas (0.9, 0.999) and eps = 1e-8.

pub fn adamw(learning_rate: impl Into<LearningRate>) -> Self

AdamW, as MLX’s optimizers.AdamW: Adam with decoupled weight decay (default 0.01).

pub fn betas(self, b1: f64, b2: f64) -> Self

Sets the betas.

pub fn weight_decay(self, weight_decay: f64) -> Self

Sets the decoupled weight decay (AdamW).

pub fn bias_correction(self, enabled: bool) -> Self

Enables bias correction.

Trait Implementations

impl Clone for Adam

fn clone(&self) -> Self

Returns a duplicate of the value.

fn clone_from(&mut self, source: &Self)

Performs copy-assignment from source.

impl Debug for Adam

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

Formats the value using the given formatter.

impl Optimizer for Adam

fn apply( &mut self, parameters: &[Array], gradients: &[Array], ) -> Result<Vec<Array>>

One step: the updated parameters.

fn update<M: Module>( &mut self, model: &mut M, gradients: &[Array], ) -> Result<()>

Applies one step to model: gradients are in the order of Module::trainable_parameters.