开发 | 如何利用微信监管你的TF训练

AI科技评论按：本文作者Coldwings，AI科技评论获其授权发布。

之前回答问题【在机器学习模型的训练期间，大概几十分钟到几小时不等，大家都会在等实验的时候做什么？】的时候，说到可以用微信来管着训练，完全不用守着。没想到这么受欢迎……

原问题下的回答如下

不知道有哪些朋友是在TF/keras/chainer/mxnet等框架下用python撸的….…

这可是python啊……上itchat，弄个微信号加自己为好友（或者自己发自己），训练进展跟着一路发消息给自己就好了，做了可视化的话顺便把图也一并发过来。

然后就能安心睡觉/逛街/泡妞/写答案了。

讲道理，甚至简单的参数调整都可以照着用手机来……

大体效果如下

当然可以做得更全面一些。最可靠的办法自然是干脆地做一个http服务或者一个rpc，然而这样往往太麻烦。本着简单高效的原则，几行代码能起到效果方便自己当然是最好的，接入微信或者web真就是不错的选择了。只是查看的话，TensorBoard就很好，但是如果想加入一些自定义操作，还是自行定制的。echat.js做成web，或者itchat做个微信服务，都是挺不赖的选择。

正文如下

这里折腾一个例子。以TensorFlow的example中，利用CNN处理MNIST的程序为例，我们做一点点小小的修改。

首先这里放上写完的代码：

#!/usr/bin/env python # coding: utf-8 ''' A Convolutional Network implementation example using TensorFlow library. This example is using the MNIST database of handwritten digits (http://yann.lecun.com/exdb/mnist/) Author: Aymeric Damien Project: https://github.com/aymericdamien/TensorFlow-Examples/ Add a itchat controller with multi thread ''' from __future__ import print_function import tensorflow as tf # Import MNIST data from tensorflow.examples.tutorials.mnist import input_data # Import itchat & threading import itchat import threading # Create a running status flag lock = threading.Lock() running = False # Parameters learning_rate = 0.001 training_iters = 200000 batch_size = 128 display_step = 10 def nn_train(wechat_name, param): global lock, running # Lock with lock: running = True # mnist data reading mnist = input_data.read_data_sets("data/", one_hot=True) # Parameters # learning_rate = 0.001 # training_iters = 200000 # batch_size = 128 # display_step = 10 learning_rate, training_iters, batch_size, display_step = param # Network Parameters n_input = 784 # MNIST data input (img shape: 28*28) n_classes = 10 # MNIST total classes (0-9 digits) dropout = 0.75 # Dropout, probability to keep units # tf Graph input x = tf.placeholder(tf.float32, [None, n_input]) y = tf.placeholder(tf.float32, [None, n_classes]) keep_prob = tf.placeholder(tf.float32) #dropout (keep probability) # Create some wrappers for simplicity def conv2d(x, W, b, strides=1): # Conv2D wrapper, with bias and relu activation x = tf.nn.conv2d(x, W, strides=[1, strides, strides, 1], padding='SAME') x = tf.nn.bias_add(x, b) return tf.nn.relu(x) def maxpool2d(x, k=2): # MaxPool2D wrapper return tf.nn.max_pool(x, ksize=[1, k, k, 1], strides=[1, k, k, 1], padding='SAME') # Create model def conv_net(x, weights, biases, dropout): # Reshape input picture x = tf.reshape(x, shape=[-1, 28, 28, 1]) # Convolution Layer conv1 = conv2d(x, weights['wc1'], biases['bc1']) # Max Pooling (down-sampling) conv1 = maxpool2d(conv1, k=2) # Convolution Layer conv2 = conv2d(conv1, weights['wc2'], biases['bc2']) # Max Pooling (down-sampling) conv2 = maxpool2d(conv2, k=2) # Fully connected layer # Reshape conv2 output to fit fully connected layer input fc1 = tf.reshape(conv2, [-1, weights['wd1'].get_shape().as_list()[0]]) fc1 = tf.add(tf.matmul(fc1, weights['wd1']), biases['bd1']) fc1 = tf.nn.relu(fc1) # Apply Dropout fc1 = tf.nn.dropout(fc1, dropout) # Output, class prediction out = tf.add(tf.matmul(fc1, weights['out']), biases['out']) return out # Store layers weight & bias weights = { # 5x5 conv, 1 input, 32 outputs 'wc1': tf.Variable(tf.random_normal([5, 5, 1, 32])), # 5x5 conv, 32 inputs, 64 outputs 'wc2': tf.Variable(tf.random_normal([5, 5, 32, 64])), # fully connected, 7*7*64 inputs, 1024 outputs 'wd1': tf.Variable(tf.random_normal([7*7*64, 1024])), # 1024 inputs, 10 outputs (class prediction) 'out': tf.Variable(tf.random_normal([1024, n_classes])) } biases = { 'bc1': tf.Variable(tf.random_normal([32])), 'bc2': tf.Variable(tf.random_normal([64])), 'bd1': tf.Variable(tf.random_normal([1024])), 'out': tf.Variable(tf.random_normal([n_classes])) } # Construct model pred = conv_net(x, weights, biases, keep_prob) # Define loss and optimizer cost = tf.reduce_mean(tf.nn.softmax_cross_entropy_with_logits(logits=pred, labels=y)) optimizer = tf.train.AdamOptimizer(learning_rate=learning_rate).minimize(cost) # Evaluate model correct_pred = tf.equal(tf.argmax(pred, 1), tf.argmax(y, 1)) accuracy = tf.reduce_mean(tf.cast(correct_pred, tf.float32)) # Initializing the variables init = tf.global_variables_initializer() # Launch the graph with tf.Session() as sess: sess.run(init) step = 1 # Keep training until reach max iterations print('Wait for lock') with lock: run_state = running print('Start') while step * batch_size < training_iters and run_state: batch_x, batch_y = mnist.train.next_batch(batch_size) # Run optimization op (backprop) sess.run(optimizer, feed_dict={x: batch_x, y: batch_y, keep_prob: dropout}) if step % display_step == 0: # Calculate batch loss and accuracy loss, acc = sess.run([cost, accuracy], feed_dict={x: batch_x, y: batch_y, keep_prob: 1.}) print("Iter " + str(step*batch_size) + ", Minibatch Loss= " + \ "{:.6f}".format(loss) + ", Training Accuracy= " + \ "{:.5f}".format(acc)) itchat.send("Iter " + str(step*batch_size) + ", Minibatch Loss= " + \ "{:.6f}".format(loss) + ", Training Accuracy= " + \ "{:.5f}".format(acc), wechat_name) step += 1 with lock: run_state = running print("Optimization Finished!") itchat.send("Optimization Finished!", wechat_name) # Calculate accuracy for 256 mnist test images print("Testing Accuracy:", \ sess.run(accuracy, feed_dict={x: mnist.test.images[:256], y: mnist.test.labels[:256], keep_prob: 1.})) itchat.send("Testing Accuracy: %s" % sess.run(accuracy, feed_dict={x: mnist.test.images[:256], y: mnist.test.labels[:256], keep_prob: 1.}), wechat_name) with lock: running = False @itchat.msg_register([itchat.content.TEXT]) def chat_trigger(msg): global lock, running, learning_rate, training_iters, batch_size, display_step if msg['Text'] == u'开始': print('Starting') with lock: run_state = running if not run_state: try: threading.Thread(target=nn_train, args=(msg['FromUserName'], (learning_rate, training_iters, batch_size, display_step))).start() except: msg.reply('Running') elif msg['Text'] == u'停止': print('Stopping') with lock: running = False elif msg['Text'] == u'参数': itchat.send('lr=%f, ti=%d, bs=%d, ds=%d'%(learning_rate, training_iters, batch_size, display_step),msg['FromUserName']) else: try: param = msg['Text'].split() key, value = param print(key, value) if key == 'lr': learning_rate = float(value) elif key == 'ti': training_iters = int(value) elif key == 'bs': batch_size = int(value) elif key == 'ds': display_step = int(value) except: pass if __name__ == '__main__': itchat.auto_login(hotReload=True) itchat.run()

这段代码里面，我所做的修改主要是：

0.导入了itchat和threading

1. 把原本的脚本里网络构成和训练的部分甩到了一个函数nn_train里

def nn_train(wechat_name, param): global lock, running # Lock with lock: running = True # mnist data reading mnist = input_data.read_data_sets("data/", one_hot=True) # Parameters # learning_rate = 0.001 # training_iters = 200000 # batch_size = 128 # display_step = 10 learning_rate, training_iters, batch_size, display_step = param # Network Parameters n_input = 784 # MNIST data input (img shape: 28*28) n_classes = 10 # MNIST total classes (0-9 digits) dropout = 0.75 # Dropout, probability to keep units # tf Graph input x = tf.placeholder(tf.float32, [None, n_input]) y = tf.placeholder(tf.float32, [None, n_classes]) keep_prob = tf.placeholder(tf.float32) #dropout (keep probability) # Create some wrappers for simplicity def conv2d(x, W, b, strides=1): # Conv2D wrapper, with bias and relu activation x = tf.nn.conv2d(x, W, strides=[1, strides, strides, 1], padding='SAME') x = tf.nn.bias_add(x, b) return tf.nn.relu(x) def maxpool2d(x, k=2): # MaxPool2D wrapper return tf.nn.max_pool(x, ksize=[1, k, k, 1], strides=[1, k, k, 1], padding='SAME') # Create model def conv_net(x, weights, biases, dropout): # Reshape input picture x = tf.reshape(x, shape=[-1, 28, 28, 1]) # Convolution Layer conv1 = conv2d(x, weights['wc1'], biases['bc1']) # Max Pooling (down-sampling) conv1 = maxpool2d(conv1, k=2) # Convolution Layer conv2 = conv2d(conv1, weights['wc2'], biases['bc2']) # Max Pooling (down-sampling) conv2 = maxpool2d(conv2, k=2) # Fully connected layer # Reshape conv2 output to fit fully connected layer input fc1 = tf.reshape(conv2, [-1, weights['wd1'].get_shape().as_list()[0]]) fc1 = tf.add(tf.matmul(fc1, weights['wd1']), biases['bd1']) fc1 = tf.nn.relu(fc1) # Apply Dropout fc1 = tf.nn.dropout(fc1, dropout) # Output, class prediction out = tf.add(tf.matmul(fc1, weights['out']), biases['out']) return out # Store layers weight & bias weights = { # 5x5 conv, 1 input, 32 outputs 'wc1': tf.Variable(tf.random_normal([5, 5, 1, 32])), # 5x5 conv, 32 inputs, 64 outputs 'wc2': tf.Variable(tf.random_normal([5, 5, 32, 64])), # fully connected, 7*7*64 inputs, 1024 outputs 'wd1': tf.Variable(tf.random_normal([7*7*64, 1024])), # 1024 inputs, 10 outputs (class prediction) 'out': tf.Variable(tf.random_normal([1024, n_classes])) } biases = { 'bc1': tf.Variable(tf.random_normal([32])), 'bc2': tf.Variable(tf.random_normal([64])), 'bd1': tf.Variable(tf.random_normal([1024])), 'out': tf.Variable(tf.random_normal([n_classes])) } # Construct model pred = conv_net(x, weights, biases, keep_prob) # Define loss and optimizer cost = tf.reduce_mean(tf.nn.softmax_cross_entropy_with_logits(logits=pred, labels=y)) optimizer = tf.train.AdamOptimizer(learning_rate=learning_rate).minimize(cost) # Evaluate model correct_pred = tf.equal(tf.argmax(pred, 1), tf.argmax(y, 1)) accuracy = tf.reduce_mean(tf.cast(correct_pred, tf.float32)) # Initializing the variables init = tf.global_variables_initializer() # Launch the graph with tf.Session() as sess: sess.run(init) step = 1 # Keep training until reach max iterations print('Wait for lock') with lock: run_state = running print('Start') while step * batch_size < training_iters and run_state: batch_x, batch_y = mnist.train.next_batch(batch_size) # Run optimization op (backprop) sess.run(optimizer, feed_dict={x: batch_x, y: batch_y, keep_prob: dropout}) if step % display_step == 0: # Calculate batch loss and accuracy loss, acc = sess.run([cost, accuracy], feed_dict={x: batch_x, y: batch_y, keep_prob: 1.}) print("Iter " + str(step*batch_size) + ", Minibatch Loss= " + \ "{:.6f}".format(loss) + ", Training Accuracy= " + \ "{:.5f}".format(acc)) itchat.send("Iter " + str(step*batch_size) + ", Minibatch Loss= " + \ "{:.6f}".format(loss) + ", Training Accuracy= " + \ "{:.5f}".format(acc), wechat_name) step += 1 with lock: run_state = running print("Optimization Finished!") itchat.send("Optimization Finished!", wechat_name) # Calculate accuracy for 256 mnist test images print("Testing Accuracy:", \ sess.run(accuracy, feed_dict={x: mnist.test.images[:256], y: mnist.test.labels[:256], keep_prob: 1.})) itchat.send("Testing Accuracy: %s" % sess.run(accuracy, feed_dict={x: mnist.test.images[:256], y: mnist.test.labels[:256], keep_prob: 1.}), wechat_name) with lock: running = False

这里大部分是跟原本的代码一样的，不过首先所有print的地方都加了个itchat.send来输出日志，此外加了个带锁的状态量running用来做运行开关。此外，部分参数是通过函数参数传入的。

然后呢，写了个itchat的handler

@itchat.msg_register([itchat.content.TEXT]) def chat_trigger(msg): global lock, running, learning_rate, training_iters, batch_size, display_step if msg['Text'] == u'开始': print('Starting') with lock: run_state = running if not run_state: try: threading.Thread(target=nn_train, args=(msg['FromUserName'], (learning_rate, training_iters, batch_size, display_step))).start() except: msg.reply('Running')

作用是，如果收到微信消息，内容为『开始』，那就跑训练的函数（当然，为了防止阻塞，放在了另一个线程里）

最后再在脚本主流程里使用itchat登录微信并且启动itchat的服务，这样就实现了基本的控制。

if __name__ == '__main__': itchat.auto_login(hotReload=True) itchat.run()

但是我们不满足于此，我还希望可以对流程进行一些控制，对参数进行一些修改，于是乎：

@itchat.msg_register([itchat.content.TEXT]) def chat_trigger(msg): global lock, running, learning_rate, training_iters, batch_size, display_step if msg['Text'] == u'开始': print('Starting') with lock: run_state = running if not run_state: try: threading.Thread(target=nn_train, args=(msg['FromUserName'], (learning_rate, training_iters, batch_size, display_step))).start() except: msg.reply('Running') elif msg['Text'] == u'停止': print('Stopping') with lock: running = False elif msg['Text'] == u'参数': itchat.send('lr=%f, ti=%d, bs=%d, ds=%d'%(learning_rate, training_iters, batch_size, display_step),msg['FromUserName']) else: try: param = msg['Text'].split() key, value = param print(key, value) if key == 'lr': learning_rate = float(value) elif key == 'ti': training_iters = int(value) elif key == 'bs': batch_size = int(value) elif key == 'ds': display_step = int(value) except: pass

通过这个，我们可以在epoch中途停止（因为nn_train里通过检查running标志来确定是否需要停下来），也可以在训练开始前调整learning_rate等几个参数。

实在是很简单……