TensorLayer

Why we included this project

TensorLayer is a TensorFlow library with a large collection of customizable neural layers, aimed at people who want to build deep learning or reinforcement learning models without dropping to low-level TensorFlow ops. Reinforcement learning is where it stands out: the library bundles ready-made pieces for agents like DQN and A3C, and the project includes a solid set of example tutorials that show how to use them in readable code. That makes it a comfortable starting point for prototyping new model designs, and the Apache 2.0 license keeps commercial embedding simple. It received the ACM Multimedia Society's Best Open Source Software award in 2017, a useful sign the project was taken seriously. One caution for new users: the README now points to TensorLayerX as the successor, so check that project if you want active development rather than this older repo.