crowdAI / marLo

Multi Agent Reinforcement Learning using MalmÖ
MIT License
244 stars 46 forks source link

info variable is empty #74

Closed martinv closed 5 years ago

martinv commented 5 years ago

I installed Marlo following the installation instructions:

`conda create python=3.6 --name marlo conda config --add channels conda-forge conda activate marlo conda install -c crowdai malmo pip install -U marlo

python -c "import marlo" python -c "from marlo import MalmoPython"`

Then I ran the default bot:

`import marlo client_pool = [('127.0.0.1', 10000)] join_tokens = marlo.make('MarLo-FindTheGoal-v0', params={ "client_pool": client_pool })

assert len(join_tokens) == 1 join_token = join_tokens[0]

env = marlo.init(join_token)

observation = env.reset()

done = False while not done: _action = env.action_space.sample() obs, reward, done, info = env.step(_action) print("reward:", reward) print("done:", done) print("info", info) env.close() ` and my output looks like this:

reward: 0 done: False info {} reward: -0.01 done: False info {} reward: -0.02 done: False info {} reward: -0.01 done: False info {} reward: -0.01 done: False info {}

and so on.

Could you please help me fix this.

AndKram commented 5 years ago

Using info for the marlo competition is not allowed so there is a new suppress_info argument in the params passed to init that default to False. Setting that to True will give you info observations but not when submitting for evaluation in the competition.

martinv commented 5 years ago

I was using suppress_info, but unfortunately misspelled it as supress_info. The bot still ran, but the argument was ignored. I'm sorry for the false alarm.