Struct MAML 

pub struct MAML { /* private fields */ }

MAML (Model-Agnostic Meta-Learning) implementation.

MAML optimizes for a model initialization that can quickly adapt to new tasks through a few gradient steps.

Implementations

impl MAML

pub fn new(config: MAMLConfig) -> Self

Create a new MAML meta-learner.

Trait Implementations

impl Clone for MAML

fn clone(&self) -> MAML

Returns a duplicate of the value.

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

Performs copy-assignment from source.

impl Debug for MAML

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

Formats the value using the given formatter.

impl Default for MAML

fn default() -> Self

Returns the “default value” for a type.

impl MetaLearner for MAML

fn meta_step( &self, tasks: &[MetaTask], parameters: &HashMap<String, Array1<f64>>, ) -> TrainResult<(HashMap<String, Array1<f64>>, f64)>

Perform one step of meta-training on a batch of tasks.

fn adapt( &self, task: &MetaTask, parameters: &HashMap<String, Array1<f64>>, ) -> TrainResult<HashMap<String, Array1<f64>>>

Adapt parameters to a specific task (inner loop).