损失函数返回nan tensorflow [英] loss function is returning nan tensorflow

问题描述

我在这里编写了一个简单的tensorflow程序，该程序读取特征列表并尝试预测类.

 ，其中tf.Session()为sess:sess.run(tf.initialize_all_variables())范围内的纪元(hm_epochs):epoch_loss = 0itere = int(X_train.shape [0]/batch_size)最后= 0加= 1在范围内开始(迭代):x_train_epoch = X_train [last:(((start + add)* batch_size),:]y_train_epoch = y_1Hot_train.eval()[最后:((开始+添加)* batch_size),:]#print("x的形状"，x_train_epoch.shape，"y的形状"，y_train_epoch.shape)_，c = sess.run([优化器，成本]，feed_dict = {x:x_train_epoch，y:y_train_epoch})epoch_loss + = c最后=开始* batch_size加= 0print('Epoch'，epoch，'completed out of'，hm_epochs，'loss'，epoch_loss)正确= tf.equal(tf.argmax(预测，1)，tf.argmax(y，1))精度= tf.reduce_mean(tf.cast(正确，'float'))print('Accuracy:'，precision.eval({x:X_test，y:y_1Hot_test.eval()})) 

链接: https://gist.github.com/makark/79af6ca53ca27d51abb1d87c9b9bac07

数据文件: https://gist.github.com/makark/eb859f50237edb9343f3ca32aeb3be2b

但是，当我运行我的代码时，我总是感到难".我不确定发生了什么...任何帮助将不胜感激！

 警告:tensorflow:从< ipython-input-149-0164f4af7d52>:46:不推荐使用initialize_all_variables(来自tensorflow.python.ops.variables)，并将在2017-03-02之后删除.更新说明:改用`tf.global_variables_initializer`.纪元0完成了10损失nan时代1完成了10损失nan时代2完成了10输nan时代3完成了10输nan时代4完成了10次亏损nan时代5完成了10失落nan时代6完成了10失落nan时代7完成了10失落nan时代8完成了10输nan时代9完成了10失落nan精度:0.589097 

解决方案

• 输入具有nan，请通过 X [np.isnan(X)] = 0 对其进行修复.

• 输入未缩放，请使用sklearn的 StandardScaler 标准化输入.

• 使用random_normal中的stddev将权重设置为较小的初始值.

• 修复输出计算中的错误: output = tf.add(tf.matmul(l3，output_layer ['weights'])，output_layer ['biases']).

I've written a simple tensorflow program here that reads in a feature list and tries to predict the class.

with tf.Session() as sess:
        sess.run(tf.initialize_all_variables())

        for epoch in range (hm_epochs):
            epoch_loss = 0
            itere = int(X_train.shape[0]/batch_size)
            last = 0
            add = 1
            for start in range(itere):
                x_train_epoch = X_train[last: ((start + add) * batch_size),:]
                y_train_epoch = y_1Hot_train.eval()[last: ((start + add) * batch_size),:]
#                 print("shape of x", x_train_epoch.shape, "shape of y", y_train_epoch.shape)
                _, c = sess.run([optimizer, cost], feed_dict = {x: x_train_epoch, y: y_train_epoch})
                epoch_loss += c
                last = start * batch_size
                add = 0
            print('Epoch', epoch, 'completed out of', hm_epochs, 'loss', epoch_loss )
        correct = tf.equal(tf.argmax(prediction, 1), tf.argmax(y, 1))
        accuracy = tf.reduce_mean(tf.cast(correct, 'float'))
        print('Accuracy:', accuracy.eval( {x: X_test, y: y_1Hot_test.eval() }))

Link: https://gist.github.com/makark/79af6ca53ca27d51abb1d87c9b9bac07

Data file: https://gist.github.com/makark/eb859f50237edb9343f3ca32aeb3be2b

However, when I run my code I keep getting a loss of "nan". I'm not sure what is going on... any help would be much appreciated!

WARNING:tensorflow:From <ipython-input-149-0164f4af7d52>:46: initialize_all_variables (from tensorflow.python.ops.variables) is deprecated and will be removed after 2017-03-02.
Instructions for updating:
Use `tf.global_variables_initializer` instead.
Epoch 0 completed out of 10 loss nan
Epoch 1 completed out of 10 loss nan
Epoch 2 completed out of 10 loss nan
Epoch 3 completed out of 10 loss nan
Epoch 4 completed out of 10 loss nan
Epoch 5 completed out of 10 loss nan
Epoch 6 completed out of 10 loss nan
Epoch 7 completed out of 10 loss nan
Epoch 8 completed out of 10 loss nan
Epoch 9 completed out of 10 loss nan
Accuracy: 0.589097

解决方案

• the inputs have nan's, fix it by X[np.isnan(X)] = 0.

• the inputs are not scaled, use sklearn's StandardScaler to normalize your inputs.

• Set the weights to a small initial value use stddev in random_normal.

• Fix the bug in calculation of output: output = tf.add(tf.matmul(l3, output_layer['weights']),output_layer['biases'] ) .

这篇关于损失函数返回nan tensorflow的文章就介绍到这了，希望我们推荐的答案对大家有所帮助，也希望大家多多支持IT屋！