Struct NeuronPredictor 

pub struct NeuronPredictor { /* private fields */ }

A float32 Neuron predictor. Inputs, leaves, and ordered tree sums are float32; numerical thresholds round down. Device failures never select a CPU fallback.

Implementations

impl NeuronPredictor

pub fn new(model: &Model, ordinal: Option<usize>) -> Result<Self>

Validates and uploads a forest once and binds reusable, bounded row tiles.

pub fn num_trees(&self) -> usize

pub fn num_features(&self) -> usize

pub fn predict_raw( &mut self, rows: &DenseMatrix<'_>, num_trees: Option<usize>, out: &mut [f32], ) -> Result<()>

Writes one raw float32 score per row using a clamped tree prefix. Accepts f32/f64 input in either layout; f64 is quantized before decisions. Invalid dimensions, storage, or output lengths fail before any output write.

pub fn predict( &mut self, rows: &DenseMatrix<'_>, raw_score: bool, num_trees: Option<usize>, ) -> Result<Vec<f32>>

Scores rows and optionally applies the objective in host float32 arithmetic.