WeightとBiasを見るために
Kerasでモデルをsaveしたときにハマったので、一応の自分用のメモ
もし抽出で悩んだ人がいれば幸い
例として以下のMNISTの３層MLP(下はFunctional API)
# coding utf-8
from __future__ import function
import keras
from keras.datasets import mnist
from keras.models import Model
from keras.layers import Dense, Dropout, Input
from keras.optimizers import RMSprop
(x_train, y_train), (x_test, y_test) = mnist.load_data()
x_train = x_train.reshape(60000, 784)
x_test = x_test.reshape(10000, 784)
x_train = x_train.astype('float32')
x_test = x_test.astype('float32')
x_train /= 255
x_test /= 255
print(x_train.shape[0], 'train samples')
print(x_test.shape[0], 'test samples')
y_train = keras.utils.to_categorical(y_train, num_classes)
y_test = keras.utils.to_categorical(y_test, num_classes)
inputs = Input(shape=(784,))
hidden = Dense(128, activation='relu')(inputs)
predictions = Dense(10, activation='softmax')(hidden)
model = Model(inputs=inputs, outputs=predictions)
model.summary()
model.compile(loss='categorical_crossentropy',
optimizer=RMSprop(),
metrics=['accuracy'])
model.save('mlp.h5')
history = model.fit(x_train, y_train,
batch_size=batch_size,
epochs=epochs,
verbose=1)
score = model.evaluate(x_test, y_test, verbose=0)
print('Test loss:', score[0])
print('Test accuracy:', score[1])
これを実行後にmlp.h5を開いてパラメータを見てみよう
ここでいうパラメータはWeightとBiasの２つ
隠れ層(Dense)を見るために
# coding utf-8
from __future__ import print_function
from keras.models import load_model
import numpy as np
print('Loading model and parameters...')
model = load_model('mlp.h5')
model.summary()
print('Print the parameters...')
layer = model.layers[1] #summaryよりInput->[0], Dense->[1]なのでmodel.layers[1]
w = layer.get_weights()[0]
w = np.array(w)
b = layer.get_weights()[1]
b = np.array(b)
print('**Parameters shape**')
print('w.shape', w.shape)
print('b.shape', b.shape)
print('w = ', w)
print('b = ', b)
結果は、
Layer (type) Output Shape Param #
=================================================================
input_1 (InputLayer) (None, 784) 0
_________________________________________________________________
dense_1 (Dense) (None, 128) 100480
_________________________________________________________________
dense_2 (Dense) (None, 10) 1290
=================================================================
Total params: 101,770
Trainable params: 101,770
Non-trainable params: 0
_________________________________________________________________
Print the parameters...
**Parameters shape**
w.shape (784, 128)
b.shape (128,)
w = [[ 0.02914305 0.03145985 0.01455035 ... 0.06927236 -0.078566
-0.07896576]
[-0.0025496 -0.06011734 -0.04034768 ... 0.05541005 -0.06247988
0.02232351]
[-0.02663267 0.06812485 0.04543307 ... -0.06999144 -0.01722335
0.00585063]
...
[-0.00419274 -0.00560436 0.03638219 ... 0.00250916 0.06558482
0.05288389]
[-0.05047298 0.04109138 0.03568707 ... 0.01852474 -0.04672909
0.063163 ]
[-0.03907039 0.05963113 -0.07631446 ... 0.07864786 0.05980506
-0.01691122]]
b = [0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.
0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.
0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.
0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.
0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.
0. 0. 0. 0. 0. 0. 0. 0.]
Biasが息してません...
単純な話
model.save( )の位置が悪かったのが原因
model.compile( )は学習の設定、model.fit( )は学習の開始であるので、学習終わった後のデータをsaveするのが常
何も考えずに挿入した箇所が大幅に悪かったのね...
なので、
model.compile(...)
model.fit(...)
model.save(...)
上の順番に列記すればちゃんと？した数字が出ました(検証はこれから)
当たり前のことだけど、一応の自分用のメモとして書いておきます
因みに上のWeightは息しているようで、重みの初期値を割り振られただけの数値だったというオチですな
因みにの因みに、compileの前にsaveするとまた違う警告が来ちゃいます(対処法->コチラ)
皆様も私みたいなペーペーなことで泣かないように気を付けましょう...あうあう