Enum LdaLearning 

pub enum LdaLearning {
    Batch,
    Online {
        batch_size: usize,
        tau0: f64,
        kappa: f64,
        n_epochs: usize,
    },
}

How the variational parameters are updated across the corpus.

Variants

Batch

One sweep over every document per iteration, then a full replacement of lambda. Deterministic and monotone in the bound.

Online

Stochastic updates from shuffled mini-batches, with lambda moved a decaying step towards each batch’s estimate. Reaches a usable fit in far fewer passes when the corpus is large, at the cost of the monotonicity guarantee.

Fields

batch_size: usize

Documents per mini-batch.

tau0: f64

Learning-rate offset tau0, damping the first few steps.

kappa: f64

Learning-rate decay kappa, in (0.5, 1.0].

n_epochs: usize

Passes over the whole corpus.

Trait Implementations

impl Clone for LdaLearning

fn clone(&self) -> Self

Returns a duplicate of the value.

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

Performs copy-assignment from source.

impl Copy for LdaLearning

impl Debug for LdaLearning

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

Formats the value using the given formatter.

impl Default for LdaLearning

fn default() -> Self

Returns the “default value” for a type.