Struct SparseInferenceEngine 

pub struct SparseInferenceEngine { /* private fields */ }

Sparse inference engine that coordinates prediction and computation

Implementations

impl SparseInferenceEngine

pub fn new_sparse( input_dim: usize, hidden_dim: usize, sparsity_ratio: f32, ) -> Result<Self>

Create a new sparse inference engine with sparsity

The sparsity_ratio determines what fraction of neurons are kept active (0.0-1.0) e.g., sparsity_ratio=0.3 means 30% of neurons are active (70% sparsity)

pub fn new_dense(input_dim: usize, hidden_dim: usize) -> Result<Self>

Create a dense (non-sparse) inference engine for comparison

pub fn calibrate(&mut self, samples: &[Vec<f32>]) -> Result<()>

Calibrate the predictor with sample data

pub fn infer(&self, input: &[f32]) -> Result<Vec<f32>>

Run inference on an input vector

pub fn sparsity_statistics(&self) -> SparsityStats

Get sparsity statistics