def _initialize_trainers(self, trainer_config, sess):
trainer_parameters_dict = {}
self.trainers = {}
for brain_name in self.env.external_brain_names:
trainer_parameters = trainer_config['default'].copy()
if len(self.env.external_brain_names) > 1:
graph_scope = re.sub('[^0-9a-zA-Z]+', '-', brain_name)
trainer_parameters['graph_scope'] = graph_scope
trainer_parameters['summary_path'] = '{basedir}/{name}'.format(
basedir=self.summaries_dir,
name=str(self.run_id) + '_' + graph_scope)
else:
trainer_parameters['graph_scope'] = ''
trainer_parameters['summary_path'] = '{basedir}/{name}'.format(
basedir=self.summaries_dir, name=str(self.run_id))
if brain_name in trainer_config:
_brain_key = brain_name
while not isinstance(trainer_config[_brain_key], dict):
_brain_key = trainer_config[_brain_key]
for k in trainer_config[_brain_key]:
trainer_parameters[k] = trainer_config[_brain_key][k]
trainer_parameters_dict[brain_name] = trainer_parameters.copy()
for brain_name in self.env.external_brain_names:
if trainer_parameters_dict[brain_name]['trainer'] == "imitation":
self.trainers[brain_name] = BehavioralCloningTrainer(
sess, self.env, brain_name,
trainer_parameters_dict[brain_name], self.train_model,
self.seed)
elif trainer_parameters_dict[brain_name]['trainer'] == "ppo":
self.trainers[brain_name] = PPOTrainer(
sess, self.env, brain_name,
trainer_parameters_dict[brain_name], self.train_model,
self.seed)
else:
raise UnityEnvironmentException(
"The trainer config contains an unknown trainer type for brain {}"
.format(brain_name))
def _initialize_trainers(self, trainer_config, sess):
self.trainer_parameters_dict = {}
self.trainers = {}
for brain_name in self.env.external_brain_names:
trainer_parameters = trainer_config['default'].copy()
if len(self.env.external_brain_names) > 1:
graph_scope = re.sub('[^0-9a-zA-Z]+', '-', brain_name)
trainer_parameters['graph_scope'] = graph_scope
trainer_parameters['summary_path'] = '{basedir}/{name}'.format(
basedir=self.summaries_dir,
name=str(self.run_id) + '_' + graph_scope)
else:
trainer_parameters['graph_scope'] = ''
trainer_parameters['summary_path'] = '{basedir}/{name}'.format(
basedir=self.summaries_dir,
name=str(self.run_id))
if brain_name in trainer_config:
_brain_key = brain_name
while not isinstance(trainer_config[_brain_key], dict):
_brain_key = trainer_config[_brain_key]
for k in trainer_config[_brain_key]:
trainer_parameters[k] = trainer_config[_brain_key][k]
self.trainer_parameters_dict[brain_name] = trainer_parameters.copy()
for brain_name in self.env.external_brain_names:
if self.trainer_parameters_dict[brain_name]['trainer'] == "imitation":
self.trainers[brain_name] = BehavioralCloningTrainer(sess, self.env, brain_name,
self.trainer_parameters_dict[brain_name],
self.train_model, self.seed)
elif self.trainer_parameters_dict[brain_name]['trainer'] == "ppo":
self.trainers[brain_name] = PPOTrainer(sess, self.env, brain_name, self.trainer_parameters_dict[brain_name],
self.train_model, self.seed)
elif self.trainer_parameters_dict[brain_name]['trainer'] == "dqn":
self.trainers[brain_name] = DQNTrainer(sess, self.env, brain_name, self.trainer_parameters_dict[brain_name],
self.train_model, self.seed)
elif self.trainer_parameters_dict[brain_name]['trainer'] == "madqn":
self.trainers[brain_name] = MADQNTrainer(sess, self.env, brain_name, self.trainer_parameters_dict[brain_name],
self.train_model, self.seed)
elif self.trainer_parameters_dict[brain_name]['trainer'] == "mappo":
self.trainers[brain_name] = MAPPOTrainer(sess, self.env, brain_name, self.trainer_parameters_dict[brain_name],
self.train_model, self.seed)
elif self.trainer_parameters_dict[brain_name]['trainer'] == "coma":
self.trainers[brain_name] = COMATrainer(sess, self.env, brain_name, self.trainer_parameters_dict[brain_name],
self.train_model, self.seed)
else:
raise UnityEnvironmentException("The trainer config contains an unknown trainer type for brain {}"
.format(brain_name))
all_vars = tf.trainable_variables()
self.brain_vars = {}
total_vars = len(all_vars)
idx1 = 0
idx2 = int(total_vars/len(self.env.external_brain_names))
for brain_name in self.env.external_brain_names:
self.brain_vars[brain_name] = all_vars[idx1:idx2]
idx1 = idx2
idx2 = idx2*total_vars
if (self.trainer_parameters_dict[brain_name]['trainer'] == "dqn" or
self.trainer_parameters_dict[brain_name]['trainer'] == "madqn"):
self.trainers[brain_name].update_target_graph(self.brain_vars[brain_name])
if self.trainer_parameters_dict[brain_name]['trainer'] == "madqn":
if not self.trainers[brain_name].parameters['frozen']:
self.free_brain_vars = self.brain_vars[brain_name]
for brain_name in self.env.external_brain_names:
if self.trainer_parameters_dict[brain_name]['trainer'] == "madqn":
if self.trainers[brain_name].parameters['frozen']:
self.trainers[brain_name].update_frozen_brain_graph(self.brain_vars[brain_name], self.free_brain_vars)