Crate mecrab_word2vec 

mecrab-word2vec: Pure Rust Word2Vec implementation

Fast, memory-efficient word2vec training optimized for Japanese morphological analysis.

Features

• Skip-gram with negative sampling
• Multi-threaded training with Rayon
• Direct MCV1 format output
• Memory-efficient streaming

Example

use mecrab_word2vec::Word2VecBuilder;

let model = Word2VecBuilder::new()
    .vector_size(100)
    .window_size(5)
    .negative_samples(5)
    .min_count(10)
    .epochs(3)
    .threads(8)
    .build()?;

model.train_from_file("corpus.txt")?;
model.save_text("vectors.txt")?;

Structs

Vocabulary

Vocabulary with frequency counts and subsampling

Word2Vec

Word2Vec model

Word2VecBuilder

Builder for Word2Vec model

Enums

Word2VecError

Type Aliases

Result