Struct Lion 

pub struct Lion {
    pub learning_rate: LearningRate,
    pub betas: (f64, f64),
    pub weight_decay: f64,
    /* private fields */
}

Lion, as MLX’s optimizers.Lion: the sign of an interpolated momentum, with optional decoupled weight decay.

Fields

learning_rate: LearningRate

The learning rate.

betas: (f64, f64)

Interpolation (b1) and momentum (b2) rates; MLX’s default (0.9, 0.99).

weight_decay: f64

Decoupled weight decay.

Implementations

impl Lion

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

Lion with betas (0.9, 0.99) and no weight decay.

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

Sets the decoupled weight decay.

Trait Implementations

impl Clone for Lion

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 Lion

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

Formats the value using the given formatter.

impl Optimizer for Lion

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.