Skip to content
프로젝트

YaoYao1995/MEEE has 428 GitHub stars and 64 forks. GitHot measured 0 stars in 24 hours. Last checked 2026-11-18 23:39 UTC. Last push 2023-08-22.

이 수치를 측정하는 방법

측정된 스타 증가

in 24h
0
in 7 days
기록 수집 중
in 15 days
기록 수집 중
in 1 month
기록 수집 중

서로 겹치는 기간의 순증가량이며 일별 합계가 아닙니다. 측정값이 없으면 비워 둡니다. 스타는 관심도를 나타내며 품질을 보장하지 않습니다.

소개

Code to reproduce the experiments in Sample Efficient Reinforcement Learning via Model-Ensemble Exploration and Exploitation (MEEE).

학습

README.md

Original

Instruction to reproduce MEEE

Code to reproduce the experiments in Sample Efficient Reinforcement Learning via Model-Ensemble Exploration and Exploitation[abs].

It is noteworthy that our code is mainly based on MBPO, and we refer interested readers to the original code base MBPO for more details.

Installation

  1. Install MuJoCo 2.0 at ~/.mujoco/mujoco200 and copy your license key to ~/.mujoco/mjkey.txt, for example, you need to install the following dependencies first for Linux platform:
sudo yum install patchelf
sudo yum install mesa-libGL-devel mesa-libGLU-devel
sudo yum install mesa-libOSMesa-devel
sudo yum install mesa-libOSMesa
sudo yum install glfw
sudo yum install mesa-libGL
sudo yum install openmpi-devel
  1. Create a conda environment and install dependencies in requirements.txt
cd code_meee
conda create -n "your_env_name" python=3.6
conda activate "your_env_name"
# install cuda to suport tf-gpu==1.13.1
conda install cudatoolkit==10.0.130 
conda install cudnn==7.6.5
pip install -r requirements.txt

Usage

Configuration files can be found in examples/config. Use the following command to conduct experiment on Humanoid-v2:

python main.py run_local examples.development --config=examples.config.humanoid.1 --trial-gpus=1

Currently only running locally is supported, so just keep the run_local and examples.development arguments. examples.config.humanoid.1 determines the configuration file you want to use, and --trial-gpus=1 indicate that you would like to experiment with one Nvidia GPU, you could change the experiment environment and GPU used by modifying relative arguments.

Logging

The results can be found in the default directory log_dir=~/ray_meee/, you could also specify the directory in examples/config/configuration_files.

Citation

If you use this code or results in your paper, please cite our work as:

@inproceedings{yao2021sample,
  title={Sample efficient reinforcement learning via model-ensemble exploration and exploitation},
  author={Yao, Yao and Xiao, Li and An, Zhicheng and Zhang, Wanpeng and Luo, Dijun},
  booktitle={2021 IEEE International Conference on Robotics and Automation (ICRA)},
  pages={4202--4208},
  year={2021},
  organization={IEEE}
}

License

The code in this repository is released under the MIT license as found in the LICENSE file.