Module nas 

Neural Architecture Search (NAS) and AutoML algorithms.

This module provides algorithms for automated neural architecture search and hyperparameter optimization:

• search_space: DAG-based architecture search space definitions
• random_nas: Random search baseline for NAS
• evolutionary_nas: AmoebaNet-style evolutionary NAS
• differentiable: DARTS differentiable architecture search
• automl: AutoML random search for hyperparameter optimization

Re-exports

pub use automl::AutoMLConfig;

pub use automl::AutoMLOptimizer;

pub use automl::AutoMLResult;

pub use automl::HyperparamSpace;

pub use automl::HyperparamValue;

pub use differentiable::DARTSSearch;

pub use evolutionary_nas::EvolutionaryNAS;

pub use random_nas::ArchFitness;

pub use random_nas::NASResult;

pub use random_nas::ParamCountFitness;

pub use random_nas::RandomNAS;

pub use search_space::ArchEdge;

pub use search_space::ArchNode;

pub use search_space::Architecture;

pub use search_space::OpType;

pub use search_space::SearchSpace;

Modules

automl

AutoML: Automated machine learning pipeline optimisation.

differentiable

DARTS: Differentiable Architecture Search (Liu et al., ICLR 2019).

evolutionary_nas

Evolutionary Neural Architecture Search (AmoebaNet-style).

random_nas

Random Neural Architecture Search baseline.

search_space

Neural architecture search space definition.