Different Deep Reinforcement Approaches For Starcraft 2 Mini-games

In this repo we cover two algorithms for solving different starcraft mini-games.

DeepQ

One Ouput Array (current): 0-4095
Two array coordinates (under development):
Two matrixes Input, Two matrixes Output: array of [3] [group, coordenates_x, cooredantes_y]

Thanks to slamelnabarawy his repo has been a good baseline for the environemnt: Link to the repo

ikostrikov with this repo helpped us to migrate our experiments to Pytorch. Thanks to him too :)

chris-chris has with his repo has been also been an inspiration for us.

Name		Name	Last commit message	Last commit date
Latest commit History 44 Commits
models		models
sc2gym @ 4897f00		sc2gym @ 4897f00
.gitignore		.gitignore
.gitmodules		.gitmodules
README.md		README.md
collect_mineral_shards_1d_DQN.py		collect_mineral_shards_1d_DQN.py
collect_mineral_shards_1d_double_DQN.py		collect_mineral_shards_1d_double_DQN.py
collect_mineral_shards_1d_double_duel_DQN.py		collect_mineral_shards_1d_double_duel_DQN.py
collect_mineral_shards_1d_dueling_DQN.py		collect_mineral_shards_1d_dueling_DQN.py
collect_mineral_shards_DQN_2actors.py		collect_mineral_shards_DQN_2actors.py
collect_mineral_shards_DQN_2actors_DDDQN.py		collect_mineral_shards_DQN_2actors_DDDQN.py
collect_mineral_shards_a2c.py		collect_mineral_shards_a2c.py
plotGrpahs.py		plotGrpahs.py
requirement.txt		requirement.txt