Struct LearnedOptimizer 

pub struct LearnedOptimizer { /* private fields */ }

Learned optimizer.

Implementations

impl LearnedOptimizer

pub fn new() -> Self

Create a new learned optimizer with default settings.

pub fn with_strategy(self, strategy: LearningStrategy) -> Self

Set learning strategy.

pub fn with_model_type(self, model_type: ModelType) -> Self

Set model type.

pub fn with_learning_rate(self, rate: f64) -> Self

Set learning rate.

pub fn extract_features( &self, graph_desc: &HashMap<String, f64>, ) -> Result<FeatureVector, LearnedOptError>

Extract features from graph description.

pub fn observe( &mut self, features: FeatureVector, actual_cost: f64, ) -> Result<(), LearnedOptError>

Observe execution and update model (online learning).

pub fn observe_reward( &mut self, signal: RewardSignal, ) -> Result<(), LearnedOptError>

Observe reward signal for reinforcement learning.

pub fn predict_cost( &self, features: &FeatureVector, ) -> Result<CostPrediction, LearnedOptError>

Predict cost for given features.

pub fn recommend_fusion( &self, features: &FeatureVector, ) -> Result<FusionRecommendation, LearnedOptError>

Recommend whether to fuse operations.

pub fn get_stats(&self) -> &LearningStats

Get learning statistics.

pub fn evaluate_accuracy(&mut self) -> Result<f64, LearnedOptError>

Evaluate model accuracy on training data.

pub fn reset(&mut self)

Reset learning state.

Trait Implementations

impl Default for LearnedOptimizer

fn default() -> Self

Returns the “default value” for a type.