hanlp.word2vecTrain: word2vec training.

In qxde01/RHanLP: R/HanLP Interface

Description

A function of word2vec training .

Usage

hanlp.word2vecTrain(file_path = NULL, model_path = "word2vec",
  word_vectors_size = 128L, num_iters = 10L, word_min_freq = 5L,
  window_size = 5L, neg_sample = 5L, type = "skipgram",
  alpha = ifelse(type == "skipgram", 0.05, 0.025), softmax = TRUE,
  threads_num = 2L)

Arguments

file_path	Corpus file path .
model_path	Word2vec mode saving path.
word_vectors_size	Word vector size ,default 128.
num_iters	Number of iterations,default 10.
word_min_freq	This will discard words that appear less than <int> times; default is 5
window_size	Max skip length between words; default is 5.
neg_sample	Negative sample number,default 5.
type	skipgram or cbow model, default skipgram.
alpha	the starting learning rate; default is 0.025 for skipgram and 0.05 for cbow.
threads_num	the threads number default 2.

Author(s)

qxde01

Examples

## Not run: 
hanlp.word2vecTrain(file_path='text.txt',model_path = 'word2vec.txt')

## End(Not run)

qxde01/RHanLP documentation built on May 13, 2019, 8:39 a.m.