课程: Building and Deploying Deep Learning Applications with TensorFlow

今天就学习课程吧！

今天就开通帐号，24,100 门业界名师课程任您挑！

Visualize training runs

Visualize training runs - TensorFlow教程

课程: Building and Deploying Deep Learning Applications with TensorFlow

开始一个月试用 购买团队方案

Visualize training runs

“

- [Instructor] When you are building machine learning models with TensorFlow, you'll be spending a lot of time training the model and then repeating the training process with different parameters to see what works best. Using TensorBoard we can visually monitor the progress of training as it happens, and even compare different training runs against each other. Let's see how it works. First, let's open up visualize_training.py. Here we already have our computational graph defined and we have a training loop defined. And down here on line 104, we've defined the training filewriter and a testing filewriter that will write out log files during training that we can then view in TensorBoard. Let's say that we want to retrain this neural network several times, with a different number of nodes in the first layer each time. Our goal is to find out which neural network design gives us the best prediction accuracy. The problem is that each time we run the training process, additional log files…

内容