Struct TensorHistogram 

pub struct TensorHistogram {
    pub min: f32,
    pub max: f32,
    pub bins: Vec<u32>,
    pub bin_edges: Vec<f32>,
    pub total_count: u64,
}

Histogram of activation values for a single tensor.

Fields

min: f32

Observed minimum value.

max: f32

Observed maximum value.

bins: Vec<u32>

Bin counts.

bin_edges: Vec<f32>

Bin edges (len = bins.len() + 1).

total_count: u64

Total number of values collected.

Implementations

impl TensorHistogram

pub fn from_values( values: &[f32], num_bins: usize, ) -> Result<Self, CalibrationError>

Create a new histogram from a collection of f32 values.

The histogram covers the range [min, max] of the input values with num_bins equal-width bins.

pub fn merge(&mut self, other: &TensorHistogram)

Merge another histogram’s values into this one.

If the ranges differ, the histogram is rebuilt with the union range.

pub fn percentile(&self, pct: f32) -> f32

Compute the value at a given percentile (0.0 to 100.0).

Trait Implementations

impl Clone for TensorHistogram

fn clone(&self) -> TensorHistogram

Returns a duplicate of the value.

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

Performs copy-assignment from source.

impl Debug for TensorHistogram

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

Formats the value using the given formatter.