4. Load Models locally - Pommerman/pytorch-pommerman-rl GitHub Wiki
Copy the trained model from GCP server to local laptop
# run it on your local laptop
scp -C -p pommerman:/data/pytorch-pommerman-rl/trained_models/a2c/PommeFFACompetitionFast-v0.pt ./trained_models/a2c/PommeFFACompetitionFast-v0.pt
Demo the model in your local laptop
cd pytorch-pommerman-rl
python enjoy.py --env-name PommeFFACompetitionFast-v0 --load-path ./trained_models/a2c/PommeFFACompetitionFast-v0.pt
Configure the players in the code
cd pytorch-pommerman-rl/envs
# find the following code in pommerman.py and manual change the player number from 4 to 2 or vice versa
def make_env(config):
training_agent = TrainingAgent()
agent_list = [
training_agent,
pommerman.agents.SimpleAgent(),
pommerman.agents.SimpleAgent(),
pommerman.agents.SimpleAgent(),
]