Struct ModelAdapter 

pub struct ModelAdapter<T: FloatBounds> { /* private fields */ }

Model adapter for replacing layers during transfer learning

Implementations

impl<T: FloatBounds + ScalarOperand> ModelAdapter<T>

pub fn new(init_strategy: InitStrategy) -> Self

Create a new model adapter

pub fn replace_layer( &mut self, layer_name: String, new_layer: Box<dyn Layer<T>>, )

Add a layer replacement

pub fn replace_classifier( &mut self, num_classes: usize, hidden_size: usize, _layer_name: String, ) -> NeuralResult<()>

Remove the final classification layer and add a new one

pub fn apply_replacements<M>(&self, model: &mut M) -> NeuralResult<()>

where M: HasReplaceableLayer<T>,

Apply layer replacements to a model

Trait Implementations

impl<T: FloatBounds> Debug for ModelAdapter<T>

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

Formats the value using the given formatter.