Created
September 13, 2017 07:01
-
-
Save aonotas/55c9b5f9f089ba797e280ce4bd46504f to your computer and use it in GitHub Desktop.
Chainer Example for NStepLSTM code (cudnn GPU)
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
#!/usr/bin/env python | |
# -*- coding: utf-8 -*- | |
import numpy as np | |
from chainer import Variable | |
import chainer.functions as F | |
import chainer.links as L | |
from chainer import cuda | |
xp = cuda.cupy | |
# 入力データの準備 | |
x_list = [[0, 1, 2, 3], [4, 5, 6], [7, 8]] # 可変長データ (4, 3, 2)の長さのデータとする | |
x_list = [xp.array(x, dtype=xp.int32) for x in x_list] # numpyに変換する | |
n_vocab = 500 | |
emb_dim = 100 | |
word_embed=L.EmbedID(n_vocab, emb_dim, ignore_label=-1) | |
word_embed.to_gpu() | |
use_dropout = 0.25 | |
in_size = 100 | |
hidden_size = 200 | |
n_layers = 1 | |
bi_lstm=L.NStepBiLSTM(n_layers=n_layers, in_size=in_size, | |
out_size=hidden_size, dropout=use_dropout) | |
bi_lstm.to_gpu() | |
# Noneを渡すとゼロベクトルを用意してくれます. Encoder-DecoderのDecoderの時は初期ベクトルhxを渡すことが多いです. | |
hx = None | |
cx = None | |
xs_f = [] | |
for i, x in enumerate(x_list): | |
x = word_embed(Variable(x)) # Word IndexからWord Embeddingに変換 | |
x = F.dropout(x, ratio=use_dropout) | |
xs_f.append(x) | |
# xs_fのサイズは | |
# [(4, 100), (3, 100), (2, 100)]というVariableのリストになっている | |
hy, cy, ys = bi_lstm(hx=hx, cx=cx, xs=xs_f) | |
for h in ys: | |
print h.data.shape |
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment