search

BorealisAI / mtmfrl

Multi Type Mean Field Reinforcement Learning
Other
29 stars 10 forks source link

Not clear where mean action is chosen #3

Closed aravindvenu7 closed 4 years ago

aravindvenu7 commented 4 years ago

Not clear where in the code the mean action is chosen and the Boltzmann policy is updated.

Sriram94 commented 4 years ago

Mean action is updated in the scenario_battle.py file. Example, for the multibattle experiments, it is in the file mtmfrl/multibattle/mfrl/examples/battle_model/scenario_battle.py (lines 76 - 146).

Boltzmann policy is updated in base.py file in the algo folder. Example, for the multibattle experiments, it is in the file, mtmfrl/multibattle/mfrl/examples/battle_model/algo/base.py (lines 52-56).