A personal playground of different RL models and algorithms The model is written in Pytorch, with utility functions and some framework borrowed from: yandex
Python: 3.6 or above
Pytorch: 0.4.1 (CUDA)
Gym: most updated version
Just type
python main.py
Hyperparameters and games can be changed in the main.py script. A graph showing rewards during training will be save in the current directory.
Curiosity
Attention
Memory
and more...