Competition_3v3snakes
Environment
Check details in Jidi Competition RLChina2021智能体竞赛
Dependency
You need to create competition environment.
conda create -n snake1v1 python=3.6
conda activate snake1v1
pip install -r requirements.txt
How to train rl-agent
python rl_trainer/main.py
By default-parameters, the total reward of training is shown below.
You can edit different parameters, for example
python rl_trainer/main.py --algo "bicnet" --epsilon 0.8
Baseline performance:
You can locally evaluation your model.
python evaluation_local.py --my_ai rl --opponent random
How to test submission
You can locally test your submission. At Jidi platform, we evaluate your submission as same as run_log.py
Once you run this file, you can locally check battle logs in the folder named "logs".
For example,
python run_log.py --my_ai "random" --opponent "rl"
Ready to submit
- Random policy --> agent/random/submission.py
- RL policy --> all files in agent/rl/
Watch reply locally
- Open reply/reply.html in any browser.
- Load a log.
- Reply and watch ^0^.