• Tensorflow implementation output intermediate value with tf.print in function


    Tensorflow is based on static graph mode, which makes it difficult to debug when writing code. In addition to using official debugging tools, the most direct way is to output the intermediate results for viewing. However, directly using print function can only output the shape of the sensor variable, not the value. If you want […]

  • Tensorflow to print the output of each layer


    In test.py, you can directly generate a Pb file with weight through the following code, or you can convert CKPT to a Pb file through TF’s official free graph.py. constant_graph = graph_util.convert_variables_to_constants(sess, sess.graph_def,[‘net_loss/inference/encode/conv_output/conv_output’]) with tf.gfile.FastGFile(‘net_model.pb’, mode=’wb’) as f: f.write(constant_graph.SerializeToString()) In tf1.0, the output of each layer can be obtained through the Pb file with weight […]

  • An example of tensorflow drawing loss / accuracy curve


    1. multiple curves 1.1 use pyplot mode import numpy as np import matplotlib.pyplot as plt x = np.arange(1, 11, 1) plt.plot(x, x * 2, label=”First”) plt.plot(x, x * 3, label=”Second”) plt.plot(x, x * 4, label=”Third”) Plt.legend (LOC = 0, ncol = 1) ා parameter: LOC sets the display position, 0 is adaptive; ncol sets the […]

  • How to use the. Pb file of tensorflow training in production environment


    Preface Tensorflow is an open-source artificial intelligence learning system of Google. Why is it called that? Tensor means tensor, representing n-dimensional array; flow means flow, representing calculation based on data flow graph. The process of moving n-dimension numbers from one end of flow graph to the other is the process of analysis and processing by […]

  • Learning notes tf007: tensor, graph, Op, variable, placeholder, session, name scope, board comprehensive example


    The input adopts placeholder, the model receives any length vector, calculates the total output of data flow graph over time, reasonably divides the data flow graph by name scope, saves the output, accumulation and mean value of data flow graph to disk every time. [none] for any length vector, [] for scalar. The update phase […]

  • Google I / O hasn’t started yet. Let’s take a look at it first?


    Review the history of Google IO 2008-2016Do you know what I / O of Google I / O means? Let’s review the highlights of Google I / O from 2008 to 2016. Looking forward to | Google I / O 2017, can Google change the world this time?From 2008 when Google started to hold the […]

  • On tensorflow


    Installation of CUDA and cudnn You need to determine whether your computer is an n card and the supported version of the graphics card is in the following list (you can also go to the official website for details). If not, skip this step. https://developer.nvidia.com/… Https://developer.nvidia.com /… CUDA CUDA? Is a general parallel computing architecture […]

  • Tensorflow: natural style


    Neural style is a very interesting deep learning application: input a picture representing the content and a picture representing the style, and the deep learning network will output a new work integrating the style and content. Tensorflow is Google’s most popular open source deep learning framework. Author anishatalye uses tensorflow to implement the neural style […]

  • Using tensorflow from scratch to build a simple logistic regression model


    Tensorflow is a Python based machine learning framework. After learning the course content of logical regression on coursera, I want to re implement the content implemented in MATLAB with tensorflow, as a stepping stone for learning Python and framework. Target readers Know what logical regression is, know a little python, and have heard of tensorflow […]

  • Tensorflow take off tour · installation · demo


    Why write At first, I thought that it was very convenient to install tensorflow. After the installation, it took me two days. There was no other explanation except carelessness. Don’t say that my IQ was low. I knew it myself… Tensorflow is really powerful after the update. It exceeds a series of in-depth learning frameworks […]

  • Tensorflow bitcoin robot: tensorflow bitcoin Trading Robot


    brief introduction A bitcoin Trading Robot is based on tensorflow LSTM model and is only for entertainment.A Bitcoin trade robot based on Tensorflow LSTM model.Just for fun. data set The data comes from btctrade, crawling with requests, which contains 50 transactions of bitcoin.Get trade.py will get these trade records and show them in pictures. Model […]

  • Constructing multilayer perceptron neural network for text recognition of digital pictures


    Something interesting?!! Image recognition In keras environment, multi-layer perceptron model is constructed to recognize digital image accurately.The model does not consume a lot of computing resources, uses the CPU version of keras, uses tensorflow as the backend, and writes it in the jupyter notebook, the IPython interactive environment. 1. Data source Download the open source […]