def test_model_creation(self):
from mlpy.mdp import MDPModelFactory
# create discrete model
with pytest.raises(ValueError):
MDPModelFactory.create('discretemodel')
from mlpy.mdp.stateaction import Action
Action.description = {
'out': {'value': [-0.004]},
'in': {'value': [0.004]},
'kick': {'value': [-1.0]}
}
MDPModelFactory.create('discretemodel')
# create decision tree model
MDPModelFactory.create('decisiontreemodel')
MDPModelFactory.create('decisiontreemodel', explorer_type='unknownbonusexplorer')
MDPModelFactory.create('decisiontreemodel', explorer_type='leastvisitedbonusexplorer',
explorer_params={'rmax': 1.0})
with pytest.raises(ValueError):
MDPModelFactory.create('decisiontreemodel', explorer_type='undefined')
# create CASML model
case_template = {
"state": {
"type": "float",
"value": "data.state",
"is_index": True,
"retrieval_method": "radius-n",
"retrieval_method_params": 0.01
},
"delta_state": {
"type": "float",
"value": "data.next_state - data.state",
"is_index": False,
}
}
MDPModelFactory.create('casml', case_template)
demo = data["state"]
except IOError:
sys.exit(sys.exc_info()[1])
except KeyError, e:
sys.exit("Key not found: {0}".format(e))
kwargs = {}
if args.model == "decisiontreemodel":
kwargs = {"use_reward_trees": args.use_reward_trees}
if args.explorer_type in ["unvisitedbonusexplorer", "leastvisitedbonusexplorer", "unknownbonusexplorer"]:
kwargs.update({"explorer_type": args.explorer_type, "rmax": args.rmax})
if args.explorer_type == "leastvisitedbonusexplorer" and args.thresh:
kwargs.update({"thresh": args.thresh})
else:
args.ignore_unreachable = False
model = MDPModelFactory.create(args.model, **kwargs)
explorer = None
if args.explorer_type in ["egreedyexplorer", "softmaxexplorer"]:
explorer = ExplorerFactory.create(args.explorer_type, args.explorer_params, args.decay)
if args.learner == "apprenticeshiplearner":
learner = None
if args.progress:
try:
learner = ApprenticeshipLearner.load(args.savefile)
except IOError:
pass
if not learner:
try:
