TensorFlow 2/2.0 In Chinese
TensorFlow2 documentation series links to articles:
- TensorFlow 2/2.0 中文 版 (Jul 9, 2019)
- TensorFlow 2 英 文 版 – MNIST Image Classification (Jul 9, 2019)
- TensorFlow 2 英 文 版 – IMDB text Classification (Jul 9, 2019)
- TensorFlow 2 英 文 版 – Feature Engineering Structured Data Classification (Jul 9, 2019)
- TensorFlow 2 英 文 版 – Regression Prediction of Fuel Efficiency (Jul 11, 2019)
- TensorFlow 2 英 文 版 – Overfitting and Underfitting (Jul 12, 2019)
- TensorFlow 2 英 文 版 – Save and load model (Jul 13, 2019)
- TensorFlow 2 英 文 书 名 – Convolutional Neural Network Classification CIFAR-10 (Jul 19, 2019)
- TensorFlow 2 英 文文档 – TFHub migration learning (Jul 19, 2019)
- TensorFlow 2 英 文 版 – RNN LSTM Text Classification (Jul 22, 2019)
The source code/data set has been uploaded to Github-tensorflow2-docs-zh
The document address
- Document address: TensorFlow 2/2.0
- Github: Github – tensorflow2-docs
- Zhihu column: Zhihu – tensorflow2-docs
Table of Contents (continuously updated)
Basics – ML Basics for Machine learning
- Classify images
- Classify text
- Classify Structured Data
- Return to the Regression
- Overfitting and underfitting
- Save and restore models
Basics – Image classification
- Convolutional Neural Networks
- TFHub is used to learn TensorFlow Hub with Keras
- Pretrained CNN was used for Transfer Learning Using Pretrained ConvNets
Basic – Text classification
Advanced – Custom
- Sors and operations Tensors and operations
- Custom layers
- Automatic differentiation
- Custom training: Walkthrough
- Dynamic graph mechanism TF Function and AutoGraph
Geek bunny combat
Supervised learning
- Mnist Handwritten Digit Recognition (CNN Convolutional Neural Network)
- Supervise learning and playing OpenAI Gym Game
Reinforcement learning
- Q-Learning to play OpenAI Gym
- Intensive learning DQN play Gym Mountain Car
- In this Gradient, we learned 70 lines of code
The statement
TensorFlow 2 Chinese document is written by referring to the official website of TensorFlow. Selected some valuable chapters as a summary, the content of the catalog is basically consistent with the official document, but in the content of a lot of simplification, mainly code practice. TensorFlow is a high-level framework for machine learning, with powerful functions and many interfaces. TensorFlow 2 has discarded a large number of repetitive interfaces and adopted Keras as the main interface for building networks. It has also added many new features, greatly improving usability and effectively reducing the amount of code.
The purpose of TensorFlow 2 Chinese document is to select representative content from official documents to help you get a quick introduction to TensorFlow in image recognition, text classification, structured data, and other aspects. TensorFlow 1.x has been extensively documented, so this document focuses on summarizing the new features of TensorFlow 2.
TensorFlow documents comply with the Signature 4.0 International (CC BY 4.0) protocol and the Apache 2.0 protocol. The source will be noted prominently.
The code is implemented based on Python3 and TensorFlow 2.0 beta.
Some code has been deleted, for example, Python 2.x compatible code has been deleted.
This permanent domain is “geektutu.com”. You can also search “Geek Tutu” to find me.
Look forward to my “Zhihu column”, all articles can be viewed on the “Zhihu APP”.
Wechat scan, read/share
#TensorFlow 2 # Official documentation
Topic: TensorFlow2 documentation
2019-07-09 00:10:10, last modified 2019-08-03 01:56:43
This site article welcome link to share, prohibit full – text reprint.
TensorFlow 2 中文文档 – TensorFlow 2 MNIST image classification