Struct LowRankPredictor 

pub struct LowRankPredictor { /* private fields */ }

Low-rank activation predictor using P·Q factorization.

This predictor uses a low-rank approximation to predict which neurons will be active before performing the full computation:

• P matrix [r, input_dim]: Compresses input to rank r
• Q matrix [hidden_dim, r]: Scores neurons based on compressed input

The prediction process:

1. Compress input: z = P · x (r dimensions)
2. Score neurons: scores = Q · z (hidden_dim dimensions)
3. Select active neurons based on threshold or top-K

Implementations

impl LowRankPredictor

pub fn new( input_dim: usize, hidden_dim: usize, rank: usize, config: SparsityConfig, ) -> Result<Self>

Create a new low-rank predictor with random initialization.

pub fn from_matrices( p_matrix: Array2<f32>, q_matrix: Array2<f32>, config: SparsityConfig, ) -> Result<Self>

Create from existing matrices.

pub fn rank(&self) -> usize

Get the rank of the predictor.

pub fn input_dim(&self) -> usize

Get input dimension.

pub fn hidden_dim(&self) -> usize

Get hidden dimension (number of neurons).

Trait Implementations

impl Clone for LowRankPredictor

fn clone(&self) -> LowRankPredictor

Returns a duplicate of the value.

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

Performs copy-assignment from source.

impl Debug for LowRankPredictor

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

Formats the value using the given formatter.

impl<'de> Deserialize<'de> for LowRankPredictor

fn deserialize<__D>(__deserializer: __D) -> Result<Self, __D::Error>

where __D: Deserializer<'de>,

Deserialize this value from the given Serde deserializer.

impl Predictor for LowRankPredictor

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

Predict active neurons for the given input.

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

Calibrate the predictor using sample data.

fn stats(&self) -> PredictorStats

Get predictor statistics.

impl Serialize for LowRankPredictor

fn serialize<__S>(&self, __serializer: __S) -> Result<__S::Ok, __S::Error>

where __S: Serializer,

Serialize this value into the given Serde serializer.