Struct Word2Vec 

pub struct Word2Vec<'a> {
    pub input: &'a mut Matrix,
    /* private fields */
}

Initialise the word2vec model

Fields

• input: The input matrix
• output: The output matrix
• dim: The dimension of the model
• lr: The learning rate
• neg: The number of negative samples
• grad: The gradient vector
• neg_pos: The number of negative samples per positive sample
• sigmoid_table: The sigmoid table
• log_table: The log table
• negative_table: The negative table
• loss: The loss
• n_samples: The number of samples

Fields

input: &'a mut Matrix

Implementations

impl<'a> Word2Vec<'a>

pub fn new( input: &'a mut Matrix, output: &'a mut Matrix, dim: usize, lr: f32, neg: usize, neg_table: Arc<Vec<usize>>, neg_start: usize, ) -> Word2Vec<'a>

Generate a new word2vec model

Params

• input - The input matrix
• output - The output matrix
• dim - The dimension of the model
• lr - The learning rate
• neg - The number of negative samples
• neg_table - The negative table

Returns

A new initialised word2vec model

pub fn get_loss(&self) -> f64

Return the loss value

Returns

The loss value

pub fn set_lr(&mut self, lr: f32)

Set the learning rate

Params

• lr - The learning rate

pub fn get_lr(&self) -> f32

Get the learning rate

Returns

The learning rate

pub fn update(&mut self, input: usize, target: usize)

Update the model

Params

• input - The input value
• target - The target value