分析：
看 TensorFlow 也有一段时间了，准备按照 GitHub 上的教程，敲出来，顺便整理一下思路。
RNN部分
- 定义参数，包括数据相关，训练相关。
- 定义模型，损失函数，优化函数。
- 训练，准备数据，输入数据，输出结果。
代码：
#!/usr/bin/env python
# -*- coding: utf-8 -*-
import tensorflow as tf
from tensorflow.examples.tutorials.mnist import input_data
from tensorflow.contrib import rnn
mnist=input_data.read_data_sets("./data",one_hot=True)
training_rate=0.001
training_iters=100000
batch_size=128
display_step=10
n_input=28
n_steps=28
n_hidden=128
n_classes=10
x=tf.placeholder("float",[None,n_steps,n_input])
y=tf.placeholder("float",[None,n_classes])
weights={'out':tf.Variable(tf.random_normal([n_hidden,n_classes]))}
biases={'out':tf.Variable(tf.random_normal([n_classes]))}
def RNN(x,weights,biases):
x=tf.unstack(x,n_steps,1)
lstm_cell=rnn.BasicLSTMCell(n_hidden,forget_bias=1.0)
outputs,states=rnn.static_rnn(lstm_cell,x,dtype=tf.float32)
return tf.matmul(outputs[-1],weights['out'])+biases['out']
pred=RNN(x,weights,biases)
cost=tf.reduce_mean(tf.nn.softmax_cross_entropy_with_logits(logits=pred,labels=y))
optimizer=tf.train.AdamOptimizer(learning_rate=training_rate).minimize(cost)
correct_pred=tf.equal(tf.argmax(pred,1),tf.argmax(y,1))
accuaracy=tf.reduce_mean(tf.cast(correct_pred,tf.float32))
init=tf.global_variables_initializer()
with tf.Session() as sess:
sess.run(init)
step=1
while step*batch_size<training_iters:
batch_x,batch_y=mnist.train.next_batch(batch_size)
batch_x=batch_x.reshape(batch_size,n_steps,n_input)
sess.run(optimizer,feed_dict={x:batch_x,y:batch_y})
if step%display_step==0:
acc=sess.run(accuaracy,feed_dict={x:batch_x,y:batch_y})
loss = sess.run(cost, feed_dict={x: batch_x, y: batch_y})
print("Iter " + str(step * batch_size) + ", Minibatch Loss= " +
"{:.6f}".format(loss) + ", Training Accuracy= " +
"{:.5f}".format(acc))
step+=1输出：
/anaconda/bin/python2.7 /Users/xxxx/PycharmProjects/TF_3/tf_rnn.py
Extracting ./data/train-images-idx3-ubyte.gz
Extracting ./data/train-labels-idx1-ubyte.gz
Extracting ./data/t10k-images-idx3-ubyte.gz
Extracting ./data/t10k-labels-idx1-ubyte.gz
2017-07-15 16:41:15.125981: W tensorflow/core/platform/cpu_feature_guard.cc:45] The TensorFlow library wasn’t compiled to use SSE4.2 instructions, but these are available on your machine and could speed up CPU computations.
2017-07-15 16:41:15.125994: W tensorflow/core/platform/cpu_feature_guard.cc:45] The TensorFlow library wasn’t compiled to use AVX instructions, but these are available on your machine and could speed up CPU computations.
2017-07-15 16:41:15.125997: W tensorflow/core/platform/cpu_feature_guard.cc:45] The TensorFlow library wasn’t compiled to use AVX2 instructions, but these are available on your machine and could speed up CPU computations.
2017-07-15 16:41:15.126002: W tensorflow/core/platform/cpu_feature_guard.cc:45] The TensorFlow library wasn’t compiled to use FMA instructions, but these are available on your machine and could speed up CPU computations.
Iter 1280, Minibatch Loss= 1.842738, Training Accuracy= 0.33594
Iter 2560, Minibatch Loss= 1.489123, Training Accuracy= 0.50000
Iter 3840, Minibatch Loss= 1.300060, Training Accuracy= 0.57812
Iter 5120, Minibatch Loss= 1.244872, Training Accuracy= 0.62500
Iter 6400, Minibatch Loss= 0.947143, Training Accuracy= 0.71094
Iter 7680, Minibatch Loss= 0.709695, Training Accuracy= 0.75781
Iter 8960, Minibatch Loss= 0.799844, Training Accuracy= 0.76562
Iter 10240, Minibatch Loss= 0.594611, Training Accuracy= 0.83594
Iter 11520, Minibatch Loss= 0.529350, Training Accuracy= 0.82031
Iter 12800, Minibatch Loss= 0.624426, Training Accuracy= 0.82031
Iter 14080, Minibatch Loss= 0.481889, Training Accuracy= 0.82812
Iter 15360, Minibatch Loss= 0.449692, Training Accuracy= 0.84375
Iter 16640, Minibatch Loss= 0.418820, Training Accuracy= 0.85938
Iter 17920, Minibatch Loss= 0.412161, Training Accuracy= 0.85156
Iter 19200, Minibatch Loss= 0.256099, Training Accuracy= 0.90625
Iter 20480, Minibatch Loss= 0.227309, Training Accuracy= 0.90625
Iter 21760, Minibatch Loss= 0.431014, Training Accuracy= 0.85938
Iter 23040, Minibatch Loss= 0.377097, Training Accuracy= 0.87500
Iter 24320, Minibatch Loss= 0.268153, Training Accuracy= 0.89844
Iter 25600, Minibatch Loss= 0.170557, Training Accuracy= 0.95312
Iter 26880, Minibatch Loss= 0.286947, Training Accuracy= 0.91406
Iter 28160, Minibatch Loss= 0.189623, Training Accuracy= 0.94531
Iter 29440, Minibatch Loss= 0.228949, Training Accuracy= 0.95312
Iter 30720, Minibatch Loss= 0.157198, Training Accuracy= 0.94531
Iter 32000, Minibatch Loss= 0.205744, Training Accuracy= 0.93750
Iter 33280, Minibatch Loss= 0.195218, Training Accuracy= 0.92188
Iter 34560, Minibatch Loss= 0.177956, Training Accuracy= 0.92969
Iter 35840, Minibatch Loss= 0.131563, Training Accuracy= 0.96875
Iter 37120, Minibatch Loss= 0.215156, Training Accuracy= 0.92969
Iter 38400, Minibatch Loss= 0.232274, Training Accuracy= 0.94531
Iter 39680, Minibatch Loss= 0.324053, Training Accuracy= 0.91406
Iter 40960, Minibatch Loss= 0.196385, Training Accuracy= 0.93750
Iter 42240, Minibatch Loss= 0.151221, Training Accuracy= 0.95312
Iter 43520, Minibatch Loss= 0.242021, Training Accuracy= 0.95312
Iter 44800, Minibatch Loss= 0.304008, Training Accuracy= 0.90625
Iter 46080, Minibatch Loss= 0.185177, Training Accuracy= 0.93750
Iter 47360, Minibatch Loss= 0.190960, Training Accuracy= 0.94531
Iter 48640, Minibatch Loss= 0.141995, Training Accuracy= 0.94531
Iter 49920, Minibatch Loss= 0.199995, Training Accuracy= 0.94531
Iter 51200, Minibatch Loss= 0.193773, Training Accuracy= 0.92188
Iter 52480, Minibatch Loss= 0.151757, Training Accuracy= 0.94531
Iter 53760, Minibatch Loss= 0.153755, Training Accuracy= 0.94531
Iter 55040, Minibatch Loss= 0.141472, Training Accuracy= 0.93750
Iter 56320, Minibatch Loss= 0.168057, Training Accuracy= 0.96094
Iter 57600, Minibatch Loss= 0.135691, Training Accuracy= 0.96094
Iter 58880, Minibatch Loss= 0.097003, Training Accuracy= 0.97656
Iter 60160, Minibatch Loss= 0.274090, Training Accuracy= 0.92188
Iter 61440, Minibatch Loss= 0.147230, Training Accuracy= 0.95312
Iter 62720, Minibatch Loss= 0.106019, Training Accuracy= 0.96094
Iter 64000, Minibatch Loss= 0.101133, Training Accuracy= 0.97656
Iter 65280, Minibatch Loss= 0.169548, Training Accuracy= 0.93750
Iter 66560, Minibatch Loss= 0.101966, Training Accuracy= 0.96094
Iter 67840, Minibatch Loss= 0.106501, Training Accuracy= 0.96875
Iter 69120, Minibatch Loss= 0.082817, Training Accuracy= 0.96875
Iter 70400, Minibatch Loss= 0.192926, Training Accuracy= 0.96094
Iter 71680, Minibatch Loss= 0.086935, Training Accuracy= 0.96875
Iter 72960, Minibatch Loss= 0.052052, Training Accuracy= 0.98438
Iter 74240, Minibatch Loss= 0.129968, Training Accuracy= 0.95312
Iter 75520, Minibatch Loss= 0.058070, Training Accuracy= 0.99219
Iter 76800, Minibatch Loss= 0.089518, Training Accuracy= 0.96875
Iter 78080, Minibatch Loss= 0.106092, Training Accuracy= 0.98438
Iter 79360, Minibatch Loss= 0.223101, Training Accuracy= 0.92188
Iter 80640, Minibatch Loss= 0.069419, Training Accuracy= 0.97656
Iter 81920, Minibatch Loss= 0.050585, Training Accuracy= 0.99219
Iter 83200, Minibatch Loss= 0.048002, Training Accuracy= 0.98438
Iter 84480, Minibatch Loss= 0.094293, Training Accuracy= 0.96875
Iter 85760, Minibatch Loss= 0.152253, Training Accuracy= 0.96094
Iter 87040, Minibatch Loss= 0.085382, Training Accuracy= 0.97656
Iter 88320, Minibatch Loss= 0.147018, Training Accuracy= 0.95312
Iter 89600, Minibatch Loss= 0.099780, Training Accuracy= 0.96094
Iter 90880, Minibatch Loss= 0.118362, Training Accuracy= 0.93750
Iter 92160, Minibatch Loss= 0.110498, Training Accuracy= 0.96094
Iter 93440, Minibatch Loss= 0.077664, Training Accuracy= 0.98438
Iter 94720, Minibatch Loss= 0.070865, Training Accuracy= 0.96094
Iter 96000, Minibatch Loss= 0.156309, Training Accuracy= 0.94531
Iter 97280, Minibatch Loss= 0.116825, Training Accuracy= 0.94531
Iter 98560, Minibatch Loss= 0.099852, Training Accuracy= 0.96875
Iter 99840, Minibatch Loss= 0.116358, Training Accuracy= 0.96875
Process finished with exit code 0原文链接：http://www.tensorflownews.com/2017/07/15/tensorflow-rnn-turorial-mnist-code/