Tensorflow，如何存储变量？

Question

所以我刚刚开始用 tensorflow 做一些实验，但我觉得我很难掌握这个概念，我目前专注于 MNIST 数据集，但只有 8000 个用作训练，2000 个用于测试。我目前拥有的小代码片段是：

from keras.layers import Input, Dense, initializers

from keras.models import Model

from Dataset import Dataset

import matplotlib.pyplot as plt

from keras import optimizers, losses

import tensorflow as tf

import keras.backend as K

#global variables

d = Dataset()

num_features = d.X_train.shape[1]

low_dim = 32

def autoencoder():

    w = initializers.RandomNormal(mean=0.0, stddev=0.05, seed=None)

    input = Input(shape=(num_features,))

    encoded = Dense(low_dim, activation='relu', kernel_initializer = w)(input)

    decoded = Dense(num_features, activation='sigmoid', kernel_initializer = w)(encoded)

    autoencoder = Model(input, decoded)

    adam = optimizers.Adagrad(lr=0.01, epsilon=None, decay=0.0)

    autoencoder.compile(optimizer=adam, loss='binary_crossentropy')

    autoencoder.fit(d.X_train, d.X_train,

                    epochs=50,

                    batch_size=64,

                    shuffle=True,

                    )

    encoded_imgs = autoencoder.predict(d.X_test)

    decoded_imgs = autoencoder.predict(encoded_imgs)

    #sess = tf.InteractiveSession()

    #error = losses.mean_absolute_error(decoded_imgs[0], d.X_train[0])

    #print(error.eval())

    #print(decoded_imgs.shape)

    #sess.close()

    n = 20  # how many digits we will display

    plt.figure(figsize=(20, 4))

    for i in range(n):

        # display original

        #sess = tf.InteractiveSession()

        error = losses.mean_absolute_error(decoded_imgs[n], d.X_test[n])

        #print(error.eval())

        #print(decoded_imgs.shape)

        #sess.close()

        ax = plt.subplot(2, n, i + 1)

        plt.imshow(d.X_test[i].reshape(28, 28))

        plt.gray()

        ax.get_xaxis().set_visible(False)

        ax.get_yaxis().set_visible(False)

我想要做的是将错误存储为一个列表，稍后我可以将其打印或绘制在图表中，但是如何使用 tensorflow/keras 有效地做到这一点？提前致谢

Answer 1

您可以使用回调 CSVLogger 将错误存储在 csv 文件中。这是此任务的代码片段。from keras.callbacks import CSVLogger# define callbackscallbacks = [CSVLogger(path_csv_logger, separator=';', append=True)]# pass callback to model.fit() oder model.fit_generator()model.fit_generator(    train_batch, train_steps, epochs=10, callbacks=callbacks,    validation_data=validation_batch, validation_steps=val_steps)编辑：为了在列表中存储错误，你可以使用这样的东西# source https://keras.io/callbacks/class LossHistory(keras.callbacks.Callback):    def on_train_begin(self, logs={}):        self.losses = []    def on_batch_end(self, batch, logs={}):        self.losses.append(logs.get('loss'))