Struct ArchSearchSpace 

pub struct ArchSearchSpace {
    pub min_depth: usize,
    pub max_depth: usize,
    pub width_options: Vec<usize>,
    pub activation_options: Vec<String>,
    pub op_options: Vec<String>,
}

Defines the search space over neural architectures.

Constrains depth range, layer width choices, activation functions, and operation types that can appear in any sampled architecture.

Fields

min_depth: usize

Minimum number of layers (inclusive, ≥ 1).

max_depth: usize

Maximum number of layers (inclusive, ≥ min_depth).

width_options: Vec<usize>

Allowed width (hidden-unit count) options per layer.

activation_options: Vec<String>

Allowed activation function names per layer.

op_options: Vec<String>

Allowed operation type names per layer.

Implementations

impl ArchSearchSpace

pub fn new( min_depth: usize, max_depth: usize, width_options: Vec<usize>, activation_options: Vec<String>, op_options: Vec<String>, ) -> TrainResult<Self>

Construct a validated ArchSearchSpace.

Errors

Returns TrainError::InvalidParameter if:

• min_depth < 1
• max_depth < min_depth
• any of the option vecs is empty

Trait Implementations

impl Clone for ArchSearchSpace

fn clone(&self) -> ArchSearchSpace

Returns a duplicate of the value.

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

Performs copy-assignment from source.

impl Debug for ArchSearchSpace

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

Formats the value using the given formatter.