RL-mid-term

2021 Fall Reinforcement Learning Mid project

This project is modified from https://github.com/wetliu/dqn_pytorch based on my opinionated styles of coding.

Optimization method

• boost the training speed using Prioritized Experience Replay
• improve the performance using Dueling DQN

And here are a pre-trained weights that you can play with:

• model_weights_214（./models/model_weights_214）

notes：

• You can run it in the cluster jupyterLab, or install a virtual environment pytorch in the local Anconda3 environment, and run it locally.
• You need to pay attention to the version when installing Anconda3.
• Install some dependencies according to requirements.freezed.