Reinforcement-learning-with-tensorflow

Why we included this project

Morvan Zhou's tutorials work through reinforcement learning one runnable example at a time, starting with Q-learning and Sarsa in grid worlds before moving on to Deep Q-Networks and its variants (Double DQN, Prioritized Replay, Dueling DQN) and then policy-based methods like Policy Gradients, Actor-Critic, DDPG, A3C, and PPO. Each lesson pairs a short explanation with a working TensorFlow implementation and a visual environment, so you can watch the agent learn while it runs. Experiments on harder control problems, such as a 2D car, a robot arm, BipedalWalker, and LunarLander, give a sense of how these methods hold up when the task gets trickier. The material is available in both Chinese and English, which makes it a practical starting point for developers and students who learn best by reading and modifying code.