# ModelRepo

**Repository Path**: sheldonzhou/ModelRepo

## Basic Information

- **Project Name**: ModelRepo
- **Description**: reproduce some RL or Multi-Agent models
- **Primary Language**: Unknown
- **License**: Not specified
- **Default Branch**: master
- **Homepage**: None
- **GVP Project**: No

## Statistics

- **Stars**: 0
- **Forks**: 0
- **Created**: 2020-02-01
- **Last Updated**: 2020-12-19

## Categories & Tags

**Categories**: Uncategorized

**Tags**: None

## README

# ModelRepo

Reproduce (deep) RL or Multi-Agent RL models (_all of implementations are supported by gym-based (gym, particle) environments._)

## Structure

`lib` contains environments designed for (Deep) RL and multi-agent RL tasks and dependency files, e.g, `lib/ma_env` is a particle environment developed by OpenAI. All algorithms listed below are implemented independently in different sub-directory.

## Guides

**[NFSP (Neural Fictitious Self-Play)](https://github.com/KornbergFresnel/ModelRepo/tree/master/NFSP)**

NFSP is a framework for improving the performance of deep reinforcement learning tasks. You can get more details in 👇 

- arXiv link: [Deep Reinforcement Learning from Self-Play in Imperfect-Information Games](http://arxiv.org/abs/1603.01121)

- the work of implementation is still in process ...

**[Multi-Agent Deep Deterministic Policy Gradient](https://github.com/KornbergFresnel/ModelRepo/tree/master/MADDPG)**

A multi-agent deterministic policy gradient framework is proposed by *Ryan Lowe* and *Yi Wu* at 2017 which solves the non-stationary problem at training stage. Reading more by visiting 👇 arXiv link. This implementation supports gym-based multi-agent environments.

- arXiv link: [Multi-Agent Actor-Critic Mixed Cooperative-Competitive Environments](https://arxiv.org/abs/1706.02275)

- run `./scripts/run_maddpg.py` to train the model, get more information about execution: `python ./scripts/run_maddpg.py -h`

- if you wanna try different parameters configuration, you can modify the `config.py`


**[CommNet: Learning Multiagent Communication with Backpropagation](https://github.com/KornbergFresnel/ModelRepo/tree/master/CommNet)**

A simple multi-agent communication framework proposed by *Sainbayar Sukhbaatar*, *Arthur Szlam* and *Rob Fergus* at 2016.

- arXiv link: [Learning Multiagent Communication with Backpropagation](https://arxiv.org/abs/1605.07736)

- run `leaver_train.py` to train the model, you can also visit [KornbergFrsnel: CommNet](https://github.com/KornbergFresnel/CommNet) directly to get more information

- the paper has several different playgrounds, while there has only *Leaver* implemented in this repo so far (maybe I will add more playgrounds, but who knows.)

**[DDPG (CONTINUOUS CONTROL WITH DEEP REINFORCEMENT
LEARNING)](https://arxiv.org/pdf/1509.02971.pdf)**

**Independent Q-learning Network: (D)DQN / (D)DQN with Dueling**

**[SAC: Soft Actor Critic](https://arxiv.org/pdf/1801.01290.pdf)**

**[MASQ: Multiagent Soft Q-learning](http://arxiv.org/abs/1804.09817)**