self-imitation-learning-pytorch

KO | EN

gitlite — search

#hacktoberfest #typescript #android #rust #deep-learning #linux #docker #macos #java #cli #kubernetes #pytorch #openai #open-source

self-imitation-learning-pytorch

(★ 67)

This is the pytorch implementation of ICML 2018 paper - Self-Imitation Learning.

self-imitation-learning-pytorch 최신버젼 다운로드

최종 버전 다운로드 (.zip)

// repository documentation

Was this content helpful?

★ 0 (0 ratings)

# Self-Imitation-Learning with A2C
![MIT License](https://img.shields.io/badge/license-MIT-blue.svg)   
This is the pytorch version of the **A2C + SIL** - **which is basiclly the same as the openai baselines**. The paper could be found [Here](https://arxiv.org/abs/1806.05635).  
## TODO List
- [ ] Add PPO with SIL
- [ ] Add more results
## Requirements
- python-3.5.2
- openai-baselines
- pytorch-0.4.0
## Installation
Install OpenAI Baselines (**Need to use the previous version of openai-baselines, will solve in the future.**)
```bash
# clone the openai baselines
git clone https://github.com/openai/baselines.git
cd baselines
git checkout 366f486
pip install -e .

```
## How to use the code
Train the network:
```bash
python train.py --env-name 'PongNoFrameskip-v4' --cuda (if you have the GPU)

```
Test the network:
```bash
python demo.py --env-name 'PongNoFrameskip-v4'

```
You could also try the A2C algorithm without SIL by adding flag `--no-sil`:
```bash
python train.py --env-name 'PongNoFrameskip-v4' --cuda --no-sil

```
## Training Performance
Because of time, I just run Pong with 2 million steps. The results of **MontezumaRevenge** will be uploaded later!
![Scheme](https://raw.githubusercontent.com/TianhongDai/self-imitation-learning-pytorch/master/figures/sil_pong.png)  
Another results for the **Freeway** which is correspond with the original paper.  
![freeway](https://raw.githubusercontent.com/TianhongDai/self-imitation-learning-pytorch/master/figures/sil_freeway.png)
## Demo: FreewayNoFrameskip-v4
![freewaydemo](https://raw.githubusercontent.com/TianhongDai/self-imitation-learning-pytorch/master/figures/freeway.gif)
## Acknowledgement
[@junhyukoh](https://github.com/junhyukoh) for [original code](https://github.com/junhyukoh/self-imitation-learning)

self-imitation-learning-pytorch

self-imitation-learning-pytorch 최신버젼 다운로드

오늘의 추천 프로젝트

Was this content helpful?

Recent Feedback

self-imitation-learning-pytorch

self-imitation-learning-pytorch 최신버젼 다운로드

오늘의 추천 프로젝트

Was this content helpful?

Recent Feedback

Download README