「ゼロから作るDeep Learning 2 ―自然言語処理編」のRNNコード、P216、第5章「5.5.3 RNNLMの学習コード」をLSTMコードに変更する。
”「ゼロから作るDeep Learning 2 ―自然言語処理編」のRNNコードを全体が見えるようにする” で作ったコードをまんまLSTMに変更します。
実際のコードがこちら。
# ゼロから作る Deep Learning２のP216、第5章「5.5.3 RNNLMの学習コード」でコード全体が見えるようにできるだけ「import」を外したコード これをLSTM用に変更
# coding: utf-8
import sys
sys.path.append('C:\\kojin\\資料\\AI関連\\ゼロから作る Deep Learning\\ゼロから作る Deep Learning2\\deep-learning-from-scratch-2-master\\')
import matplotlib.pyplot as plt
import numpy as np
# from common.optimizer import SGD
from dataset import ptb # このimportを有効にするには上記パス設定「sys.path.append('C:\\kojin\\AI関連\\・・・」が必要！
# from simple_rnnlm import SimpleRnnlm
# ハイパーパラメータの設定
batch_size = 10
wordvec_size = 100
hidden_size = 100
time_size = 5 # Truncated BPTTの展開する時間サイズ
lr = 0.4
max_epoch = 100
# 学習データの読み込み（データセットを小さくする）
corpus, word_to_id, id_to_word = ptb.load_data('train')
corpus_size = 1000
corpus = corpus[:corpus_size]
vocab_size = int(max(corpus) + 1)
xs = corpus[:-1] # 入力
ts = corpus[1:] # 出力（教師ラベル）
data_size = len(xs)
# 学習時に使用する変数
max_iters = data_size // (batch_size * time_size)
time_idx = 0
total_loss = 0
loss_count = 0
ppl_list = []
# ///////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////
# ///////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////
# コード変更、追加箇所の「開始」部分
# ///////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////
# GPUを定義しておく（コードのどこかでこの定義を参照しているらしいけど、PCにNVIDIA無いので、下記定義をするだけ）
GPU = False
# +++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
# functions.py の抜粋「開始」部分
# ---------------------------------------------------------------------------------------------------------------------------
def sigmoid(x):
return 1 / (1 + np.exp(-x))
def softmax(x):
if x.ndim == 2:
x = x - x.max(axis=1, keepdims=True)
x = np.exp(x)
x /= x.sum(axis=1, keepdims=True)
elif x.ndim == 1:
x = x - np.max(x)
x = np.exp(x) / np.sum(np.exp(x))
return x
# ---------------------------------------------------------------------------------------------------------------------------
# functions.py の抜粋「終了」部分
# +++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
# +++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
# optimizer.py の抜粋「開始」部分
# ---------------------------------------------------------------------------------------------------------------------------
class SGD:
'''
確率的勾配降下法（Stochastic Gradient Descent）
'''
def __init__(self, lr=0.01):
self.lr = lr
def update(self, params, grads):
for i in range(len(params)):
params[i] -= self.lr * grads[i]
# ---------------------------------------------------------------------------------------------------------------------------
# optimizer.py の抜粋「終了」部分
# +++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
# +++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
# layers.py の抜粋「開始」部分
# ---------------------------------------------------------------------------------------------------------------------------
class Embedding:
def __init__(self, W):
self.params = [W]
self.grads = [np.zeros_like(W)]
self.idx = None
def forward(self, idx):
W, = self.params
self.idx = idx
out = W[idx]
return out
def backward(self, dout):
dW, = self.grads
dW[...] = 0
if GPU:
np.scatter_add(dW, self.idx, dout)
else:
np.add.at(dW, self.idx, dout)
return None
# ---------------------------------------------------------------------------------------------------------------------------
# layers.py の抜粋「終了」部分
# +++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
# +++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
# time_layers.py の抜粋「開始」部分 下記のコードは「RNN」部分を「LSTM」に置き換えたもの
# ---------------------------------------------------------------------------------------------------------------------------
class TimeEmbedding:
def __init__(self, W):
self.params = [W]
self.grads = [np.zeros_like(W)]
self.layers = None
self.W = W
def forward(self, xs):
N, T = xs.shape
V, D = self.W.shape
out = np.empty((N, T, D), dtype='f')
self.layers = []
for t in range(T):
layer = Embedding(self.W)
out[:, t, :] = layer.forward(xs[:, t])
self.layers.append(layer)
return out
def backward(self, dout):
N, T, D = dout.shape
grad = 0
for t in range(T):
layer = self.layers[t]
layer.backward(dout[:, t, :])
grad += layer.grads[0]
self.grads[0][...] = grad
return None
class TimeAffine:
def __init__(self, W, b):
self.params = [W, b]
self.grads = [np.zeros_like(W), np.zeros_like(b)]
self.x = None
def forward(self, x):
N, T, D = x.shape
W, b = self.params
rx = x.reshape(N*T, -1)
out = np.dot(rx, W) + b
self.x = x
return out.reshape(N, T, -1)
def backward(self, dout):
x = self.x
N, T, D = x.shape
W, b = self.params
dout = dout.reshape(N*T, -1)
rx = x.reshape(N*T, -1)
db = np.sum(dout, axis=0)
dW = np.dot(rx.T, dout)
dx = np.dot(dout, W.T)
dx = dx.reshape(*x.shape)
self.grads[0][...] = dW
self.grads[1][...] = db
return dx
class LSTM:
def __init__(self, Wx, Wh, b):
'''
Parameters
----------
Wx: 入力`x`用の重みパラーメタ（4つ分の重みをまとめる）
Wh: 隠れ状態`h`用の重みパラメータ（4つ分の重みをまとめる）
b: バイアス（4つ分のバイアスをまとめる）
'''
self.params = [Wx, Wh, b]
self.grads = [np.zeros_like(Wx), np.zeros_like(Wh), np.zeros_like(b)]
self.cache = None
def forward(self, x, h_prev, c_prev):
Wx, Wh, b = self.params
N, H = h_prev.shape
A = np.dot(x, Wx) + np.dot(h_prev, Wh) + b
f = A[:, :H]
g = A[:, H:2*H]
i = A[:, 2*H:3*H]
o = A[:, 3*H:]
f = sigmoid(f)
g = np.tanh(g)
i = sigmoid(i)
o = sigmoid(o)
c_next = f * c_prev + g * i
h_next = o * np.tanh(c_next)
self.cache = (x, h_prev, c_prev, i, f, g, o, c_next)
return h_next, c_next
def backward(self, dh_next, dc_next):
Wx, Wh, b = self.params
x, h_prev, c_prev, i, f, g, o, c_next = self.cache
tanh_c_next = np.tanh(c_next)
ds = dc_next + (dh_next * o) * (1 - tanh_c_next ** 2)
dc_prev = ds * f
di = ds * g
df = ds * c_prev
do = dh_next * tanh_c_next
dg = ds * i
di *= i * (1 - i)
df *= f * (1 - f)
do *= o * (1 - o)
dg *= (1 - g ** 2)
dA = np.hstack((df, dg, di, do))
dWh = np.dot(h_prev.T, dA)
dWx = np.dot(x.T, dA)
db = dA.sum(axis=0)
self.grads[0][...] = dWx
self.grads[1][...] = dWh
self.grads[2][...] = db
dx = np.dot(dA, Wx.T)
dh_prev = np.dot(dA, Wh.T)
return dx, dh_prev, dc_prev
class TimeLSTM:
def __init__(self, Wx, Wh, b, stateful=False):
self.params = [Wx, Wh, b]
self.grads = [np.zeros_like(Wx), np.zeros_like(Wh), np.zeros_like(b)]
self.layers = None
self.h, self.c = None, None
self.dh = None
self.stateful = stateful
def forward(self, xs):
Wx, Wh, b = self.params
N, T, D = xs.shape
H = Wh.shape[0]
self.layers = []
hs = np.empty((N, T, H), dtype='f')
if not self.stateful or self.h is None:
self.h = np.zeros((N, H), dtype='f')
if not self.stateful or self.c is None:
self.c = np.zeros((N, H), dtype='f')
for t in range(T):
layer = LSTM(*self.params)
self.h, self.c = layer.forward(xs[:, t, :], self.h, self.c)
hs[:, t, :] = self.h
self.layers.append(layer)
return hs
def backward(self, dhs):
Wx, Wh, b = self.params
N, T, H = dhs.shape
D = Wx.shape[0]
dxs = np.empty((N, T, D), dtype='f')
dh, dc = 0, 0
grads = [0, 0, 0]
for t in reversed(range(T)):
layer = self.layers[t]
dx, dh, dc = layer.backward(dhs[:, t, :] + dh, dc)
dxs[:, t, :] = dx
for i, grad in enumerate(layer.grads):
grads[i] += grad
for i, grad in enumerate(grads):
self.grads[i][...] = grad
self.dh = dh
return dxs
def set_state(self, h, c=None):
self.h, self.c = h, c
def reset_state(self):
self.h, self.c = None, None
class TimeSoftmaxWithLoss:
def __init__(self):
self.params, self.grads = [], []
self.cache = None
self.ignore_label = -1
def forward(self, xs, ts):
N, T, V = xs.shape
if ts.ndim == 3: # 教師ラベルがone-hotベクトルの場合
ts = ts.argmax(axis=2)
mask = (ts != self.ignore_label)
# バッチ分と時系列分をまとめる（reshape）
xs = xs.reshape(N * T, V)
ts = ts.reshape(N * T)
mask = mask.reshape(N * T)
ys = softmax(xs)
ls = np.log(ys[np.arange(N * T), ts])
ls *= mask # ignore_labelに該当するデータは損失を0にする
loss = -np.sum(ls)
loss /= mask.sum()
self.cache = (ts, ys, mask, (N, T, V))
return loss
def backward(self, dout=1):
ts, ys, mask, (N, T, V) = self.cache
dx = ys
dx[np.arange(N * T), ts] -= 1
dx *= dout
dx /= mask.sum()
dx *= mask[:, np.newaxis] # ignore_labelに該当するデータは勾配を0にする
dx = dx.reshape((N, T, V))
return dx
# ---------------------------------------------------------------------------------------------------------------------------
# time_layers.py の抜粋「終了」部分
# +++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
# +++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
# simple_rnnlm.py の抜粋「開始」部分 「LSTM」コード用に書き換えた（第6章の「rnnlm.py」からclass名以外をコピーし名前変更）
# ---------------------------------------------------------------------------------------------------------------------------
class SimpleLstmlm:
def __init__(self, vocab_size=10000, wordvec_size=100, hidden_size=100):
V, D, H = vocab_size, wordvec_size, hidden_size
rn = np.random.randn
# 重みの初期化
embed_W = (rn(V, D) / 100).astype('f')
lstm_Wx = (rn(D, 4 * H) / np.sqrt(D)).astype('f')
lstm_Wh = (rn(H, 4 * H) / np.sqrt(H)).astype('f')
lstm_b = np.zeros(4 * H).astype('f')
affine_W = (rn(H, V) / np.sqrt(H)).astype('f')
affine_b = np.zeros(V).astype('f')
# レイヤの生成
self.layers = [
TimeEmbedding(embed_W),
TimeLSTM(lstm_Wx, lstm_Wh, lstm_b, stateful=True),
TimeAffine(affine_W, affine_b)
]
self.loss_layer = TimeSoftmaxWithLoss()
self.lstm_layer = self.layers[1]
# すべての重みと勾配をリストにまとめる
self.params, self.grads = [], []
for layer in self.layers:
self.params += layer.params
self.grads += layer.grads
def predict(self, xs):
for layer in self.layers:
xs = layer.forward(xs)
return xs
def forward(self, xs, ts):
score = self.predict(xs)
loss = self.loss_layer.forward(score, ts)
return loss
def backward(self, dout=1):
dout = self.loss_layer.backward(dout)
for layer in reversed(self.layers):
dout = layer.backward(dout)
return dout
def reset_state(self):
self.lstm_layer.reset_state()
# ---------------------------------------------------------------------------------------------------------------------------
# simple_rnnlm.py の抜粋「終了」部分
# +++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
# ///////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////
# コード変更、追加箇所の「終了」部分
# ///////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////
# ///////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////
# モデルの生成
model = SimpleLstmlm(vocab_size, wordvec_size, hidden_size)
optimizer = SGD(lr)
# ミニバッチの各サンプルの読み込み開始位置を計算
jump = (corpus_size - 1) // batch_size
offsets = [i * jump for i in range(batch_size)]
for epoch in range(max_epoch):
for iter in range(max_iters):
# ミニバッチの取得
batch_x = np.empty((batch_size, time_size), dtype='i')
batch_t = np.empty((batch_size, time_size), dtype='i')
for t in range(time_size):
for i, offset in enumerate(offsets):
batch_x[i, t] = xs[(offset + time_idx) % data_size]
batch_t[i, t] = ts[(offset + time_idx) % data_size]
time_idx += 1
# 勾配を求め、パラメータを更新
loss = model.forward(batch_x, batch_t)
model.backward()
optimizer.update(model.params, model.grads)
total_loss += loss
loss_count += 1
# エポックごとにパープレキシティの評価
ppl = np.exp(total_loss / loss_count)
# print('| epoch %d | perplexity %.2f'
# % (epoch+1, ppl))
ppl_list.append(float(ppl))
total_loss, loss_count = 0, 0
# グラフの描画
x = np.arange(len(ppl_list))
plt.plot(x, ppl_list, label='train')
plt.xlabel('epochs')
plt.ylabel('perplexity')
plt.show()
実行結果がこちら。
ちゃんと動いてそうです。
RNNからLSTMへ変更するために書き換えたところは以下です。
「functions.py」から「sigmoid（）」関数を追加しました
「time_layers.py」から「class LSTM」と「class TimeLSTM」をコピーし、「class RNN」と「class TimeRNN」と入れ替えました。
「class LSTM」
と
「class TimeLSTM」
これを利用するため、「class SimpleRnnlm」の代わりに、第6章の「rnnlm.py」からclass名以外の部分をコピーし、クラス名を「class SimpleLstmlm」に変更して、以下を導入しました。
ここのところは、「ゼロから作るDeep Learning 2 ―自然言語処理編」P254、第6章「6.4 LSTMを使った言語モデル」のところのコードや解説を参照してください。
他の部分はRNNと同じなので
”「ゼロから作るDeep Learning 2 ―自然言語処理編」のRNNコードを全体が見えるようにする”
を確認してください。
以上です。
コメント
コメントを投稿