-
The exact codes used by the team "liveinparis" at the kaggle football competition
-
Implementations of self-play RL from scratch with distributed actors
-
Final version of agents currently ranked 8th/1141 at the competition (Gold prize)
-
Detailed discription of our approach
- google-research football
- PyTorch
- tensorboardX
- kaggle_environments
python3 train.py
# You can find args and hyper-parameters at the "arg_dict" in train.py.
- Orange curve - vs. easy level AI
- Blue - vs. medium level AI