This is a repository for learning series TensorFlow 2.x on every day.
The author of this repository is a machine learning enthusiast, computer science researcher, fullstack and mobile developer.
At the time of writing, the TensorFlow library has reached version 2.4.1. The way of writing code is fundamental different from tf version 1.5.x. The author of this repository intends to add as much examples of how to use different models with TensorFlow version 2.x. Starting from using Regression models in Keras, loading csv files with Panda to build a custom model with Keras and define a custom Keras Layer. It is not recommended to write model with low level tf 2.x api anymore, since the low level apis skip the model pipeline in TFX.
The author of this repository has limited time to contribute. If you have any question regarding TensorFlow 2.x, please post your question on Stack Overflow - tensorflow2.x.
Should the author of this repository has further free capacity, further tensorflow 2.x examples such as distributed training, federated learning or model training on GCP (Google Cloud Platform) might also be added. Time is always a limited factor in life besides family and work. Wish you a good time searching through the examples you might be interested in.
- 01: Using Functional API to create a custom keras model
- 02: Build a muliple output model with keras
- 03: Build a multiple input model to detect images similarity (Siamese ANN)
Follow this additional read to better understand the examples in this repository
- Beginner guide TensorFlow
- Advanced guild Keras subclassing
- Using Keras with TFX
- Train custom model with Keras: a walkthrough
- TensorBoard: TF's visualization toolkit
- Online Early Access: François Chollet, Deep Learning with Python, 2021, 2nd, Manning Publications, click on "Look Inside" to access the current MEAP.
- Online Book: Michael Nielsen, Neural Networks and Deep Learning provides a theoretical background on neural networks in general. It doesn't use tensorflow, but it is a great reference to learn more fundamentals about neural networks.
- QwikLabs - TensorFlow on GCP is an online Lab which works you trough a guided tutorial to use TensorFlow on Google Cloud Platform (GCP). You will do the Lab in a real GCP virtual machine in a time limited base. Unfortunately the QwikLab quests are not free of charge, but it total worth it to have a monthly subscription to go through some quest to gain knowledge of distributed training of tensorflow in GCP.
- TensorFlow Playground helps to understand the fundamental of neural network through a interactive fun way.
- Teachable Machine is an online Application, allows you to train simple Vison and Audio Models with your web camera or mic very quickly. You can export the models and use them for any of your school projects.