Skip to content
cimornPublic

About

A collection of implementations for deep learning and reinforcement learning algorithms, built with PyTorch.

Resources

Stars

5 stars

Watchers

0 watching

Forks

Latest commit

 

History

3 Commits

Folders and files

Repository files navigation

ML-Lab

用 PyTorch 实现机器学习和强化学习算法,记录如何运行实验以及跑出了什么结果。

目录

ML-Lab
├── README.md
├── CHANGELOG.md
├── Machine-Learning
│   ├── Regression          # 回归
│   ├── Classification      # 分类
│   └── NeuralNetwork       # 神经网络
└── Reinforcement-learning
    ├── DQN                 # 深度 Q 网络
    ├── DoubleDQN           # 双重深度 Q 网络
    ├── OPE                 # 离线策略评估
    └── PPO                 # PPO-Clip + GAE

运行实验

从仓库根目录进入你要运行的实验。例如,运行回归实验:

cd Machine-Learning/Regression
python -m pip install -r requirements.txt
python main.py

运行 OPE 时,先进入对应的环境目录。运行 PPO 时,从仓库根目录执行:

cd Reinforcement-learning/PPO
python -m pip install -r requirements.txt
python main.py train --config run/config/cartpole.json --device cpu --run-number 1

把配置换成 run/config/pendulum.json 可以运行 Pendulum 实验。有可用的 CUDA 环境时,可将设备参数改为 cuda。

实验入口

Machine-Learning

算法 实验
回归 合成数据
分类 Fashion-MNIST
神经网络 正向传播与反向传播

Reinforcement-learning

实验 任务
DQN CartPole-v1、MountainCar-v0
DoubleDQN CartPole-v1、MountainCar-v0
OPE Pendulum-v1、Swimmer-v5
PPO CartPole-v1、Pendulum-v1

实验原理与代码放在同一目录。

About

A collection of implementations for deep learning and reinforcement learning algorithms, built with PyTorch.

Resources

Stars

5 stars

Watchers

0 watching

Forks

Releases

Packages

Contributors

Languages