• first Demo Code
• Conclusion

first Demo Code

    # encoding: utf-8
    import tensorflow as tf


    tf.InteractiveSession()

the output is:

<tensorflow.python.client.session.InteractiveSession at 0x7f031ebc1dd0>

    a = tf.zeros((3,2))
    b = tf.ones((3,2))
    tf.reduce_sum(b,reduction_indices=1).eval()

the output is:

array([ 2., 2., 2.], dtype=float32)

    a.get_shape()

The output is:

TensorShape([Dimension(3), Dimension(2)])

    tf.reshape(a,(1,6)).eval()

array([[ 0., 0., 0., 0., 0., 0.]], dtype=float32)

    print type(a), type(b)

<class ‘tensorflow.python.framework.ops.Tensor’> <class ‘tensorflow.python.framework.ops.Tensor’>

Conclusion

• 需要在tensorflow的VirtualEnv中启动Ipython
• tf.InteractiveSession() 启动一个交互式的tensorflow session
• 如果需要查看一个tensor对象的value，使用该对象的eval()方法
• encoding: utf-8 =>用来解决中文乱码的问题