Experiments/Eval_RLRewards.py [161:189]:
- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -
prefix = 'downRL_pi'
increment = 20
reward_list = []

for i in range(a,b+1):

	padded_index = str(i).zfill(3)

	model_template = "{1}{0}/saved_models/Model_epoch*".format(padded_index,prefix)
	models = glob.glob(model_template)	
	# number_models = [int((model.lstrip("RL{0}/saved_models/Model_epoch".format(i))).zfill(4)) for model in models]
	# max_model = int(models[-1].lstrip("{1}{0}/saved_models/Model_epoch".format(padded_index,prefix)))
	max_model = int(remove_start(models[-1],"{1}{0}/saved_models/Model_epoch".format(padded_index,prefix)))

	max_model = max_model-max_model%increment
	model_range = np.arange(0,max_model+increment,increment)
	rewards = np.zeros((len(model_range)))

	for j in range(len(model_range)-1):
		rewards[j] = np.load("{2}{0}/MEval/m{1}/Mean_Reward_{2}{0}.npy".format(padded_index,model_range[j],prefix))
		# rewards[j] = np.load("{0}{1}/MEval/m{2}/Mean_Reward_{0}{1}.npy".format(prefix,padded_indexi,model_range[j],prefix))
	reward_list.append(rewards)

for i in range(a,b+1):

	print("For environment: ", environment_names[i-a])
	print("Average reward:", np.array(reward_list[i-a]).max())

def evalrl(a,b):
- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -



Experiments/Eval_RLRewards.py [191:219]:
- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -
	prefix = 'downRL_pi'
	increment = 20
	reward_list = []

	for i in range(a,b+1):

		padded_index = str(i).zfill(3)

		model_template = "{1}{0}/saved_models/Model_epoch*".format(padded_index,prefix)
		models = glob.glob(model_template)	
		# number_models = [int((model.lstrip("RL{0}/saved_models/Model_epoch".format(i))).zfill(4)) for model in models]
		# max_model = int(models[-1].lstrip("{1}{0}/saved_models/Model_epoch".format(padded_index,prefix)))
		max_model = int(remove_start(models[-1],"{1}{0}/saved_models/Model_epoch".format(padded_index,prefix)))

		max_model = max_model-max_model%increment
		model_range = np.arange(0,max_model+increment,increment)
		rewards = np.zeros((len(model_range)))

		for j in range(len(model_range)-1):
			rewards[j] = np.load("{2}{0}/MEval/m{1}/Mean_Reward_{2}{0}.npy".format(padded_index,model_range[j],prefix))
			# rewards[j] = np.load("{0}{1}/MEval/m{2}/Mean_Reward_{0}{1}.npy".format(prefix,padded_indexi,model_range[j],prefix))
		reward_list.append(rewards)

	for i in range(a,b+1):

		print("For environment: ", environment_names[i-a])
		print("Average reward:", np.array(reward_list[i-a]).max())

def evalrl(a,b):
- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -