Tensorflow-gpu 训练 cifar10 遇到问题

1. tensorflow 新老版本的问题，常见的问题下面的这篇博文已经列出

http://blog.csdn.net/zeuseign/article/details/72771598

2. Tensorflow cifar10_multi_gpu问题：Variable conv1/weights/ExponentialMovingAverage/ does not exist...

这个问题的解决方案是在 for i in xrange(FLAGS.num_gpus): 的外面加上with tf.variable_scope(tf.get_variable_scope()):

with tf.variable_scope(tf.get_variable_scope()):
    for i in xrange(FLAGS.num_gpus):
        with tf.device('/gpu:%d' % i):

原因：

When you do tf.get_variable_scope().reuse_variables() you set the current scope to reuse variables. If you call the optimizer in such scope, it's trying to reuse slot variables, which it cannot find, so it throws an error. If you put a scope around, the tf.get_variable_scope().reuse_variables() only affects that scope, so when you exit it, you're back in the non-reusing mode, the one you want.

参考https://stackoverflow.com/questions/41986583/tensorflow-multi-gpu-example-error-variable-conv1-weights-exponentialmovingaver

3. 可以用github上及时更新维护的tensorflow例子, 包括mnist, cifar-10, AlexNet, ImageNet

https://github.com/tensorflow/models/tree/master/tutorials/image