Skip to content

WeiChengTseng/DL_final_project

Repository files navigation

Deep Learning Final Project

Get Environemnt Binary

Please choose the binary file according to you os. If you want to rebuild the environment, please see this document.

  • MACOS
    For macos, the build binary file is in ./env/macos/SoccerTwosLearner.app
  • Ubuntu
    For Ubuntu, the build binary file is in ./env/linux/soccer_test.x86_64
  • Windows
    For Windows, the build binary file is in ./env/windows/soccer_twos

Development

Baselines

  • PPO
    • developed by Po-Han Chi
    • python main_ppo.py --folder ./PPO/{your name} --rewards_add True --reward_addtion 0.0002
    • python main_ppo.py --folder ./PPO/{your name} --rewards_add False

Model are all saved in ./PPO/

  • A2C
    • developed by Wei-Cheng Tseng
    • python main_a2c.py --env_path ./env/linux/SoccerTwosBirdView.x86_64 --reward_shaping True
    • python main_a2c.py --env_path ./env/linux/SoccerTwosBirdView.x86_64 --entropy_coeff 0.02

Multi-Agent Method

  • MADDPG
  • MAAC
    • developed by Wei-Cheng Tseng
    • python main_maac.py --env_path ./env/linux/SoccerTwosBirdView.x86_64 --steps_per_update 9600 --episode_length 300 --use_gpu
    • python main_maac.py --env_path ./env/linux/SoccerTwosBirdView.x86_64 --steps_per_update 9600 --episode_length 300 --use_gpu --buffer_length 1e5
    • python main_maac.py --env_path ./env/linux/SoccerTwosBirdView.x86_64 --steps_per_update 9600 --episode_length 300 --use_gpu --buffer_length 100000 --steps_per_update 1000 --attend_heads 2
    • python main_maac.py --env_path ./env/linux/SoccerTwosBirdView.x86_64 --steps_per_update 9600 --episode_length 300 --use_gpu --buffer_length 100000 --attend_heads 1
    • python main_maac.py --env_path ./env/linux/SoccerTwosBirdView.x86_64 --steps_per_update 9600 --episode_length 300 --use_gpu --buffer_length 100000 --attend_heads 2 --duplicate_policy True --num_updates 2
    • python main_maac_double.py --env_path ./env/linux/SoccerTwosBirdView.x86_64 --steps_per_update 9600 --episode_length 300 --use_gpu --buffer_length 100000 --attend_heads 2 --num_updates 2
    • python main_maac_ac.py --env_path ./env/linux/SoccerTwosBeta.x86_64 --steps_per_update 9600 --episode_length 300 --use_gpu --buffer_length 100000 --attend_heads 2 --num_updates 4 python main_maac_team.py --env_path ./env/linux/SoccerTwosBeta.x86_64 --steps_per_update 9600 --episode_length 300 --use_gpu --buffer_length 100000 --attend_heads 2 --num_updates 4 python main_maac_team.py --steps_per_update 9600 --episode_length 300 --use_gpu --buffer_length 100000 --attend_heads 1 --num_updates 4

Reference

About

The is the final project of deep learning in NCTU.

Resources

License

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Contributors 4

  •  
  •  
  •  
  •  

Languages