[SAC] Visual Encoder Type is not optional #3675
Description
Describe the bug
Documentation says that vis_encode_type is optional and lists a default value. However, when I omit it from my config file, I get an error:
mlagents.trainers.trainer.UnityTrainerException: The hyper-parameter vis_encode_type could not be found for the <class 'mlagents.trainers.sac.trainer.SACTrainer'> trainer of brain Helicopter.
To Reproduce
Here is my config for SAC trainer without vis_encode_type:
Helicopter:
trainer: sac
buffer_size: 4e+6
buffer_init_steps: 1000
batch_size: 1024
init_entcoef: 0.5
train_interval: 20
num_update: 1
tau: 0.005
learning_rate: 1e-3
learning_rate_schedule: linear
time_horizon: 32
max_steps: 1e6
normalize: true
num_layers: 2
hidden_units: 128
Console logs / stack traces
Traceback (most recent call last):
File "D:\Program Files\Python36\Scripts\mlagents-learn-script.py", line 11, in <module>
load_entry_point('mlagents', 'console_scripts', 'mlagents-learn')()
File "e:\my documents\unity\ml-agents\ml-agents\mlagents\trainers\learn.py", line 479, in main
run_cli(parse_command_line())
File "e:\my documents\unity\ml-agents\ml-agents\mlagents\trainers\learn.py", line 475, in run_cli
run_training(run_seed, options)
File "e:\my documents\unity\ml-agents\ml-agents\mlagents\trainers\learn.py", line 320, in run_training
tc.start_learning(env_manager)
File "e:\my documents\unity\ml-agents\ml-agents\mlagents\trainers\trainer_controller.py", line 212, in start_learning
self._create_trainers_and_managers(env_manager, new_behavior_ids)
File "e:\my documents\unity\ml-agents\ml-agents\mlagents\trainers\trainer_controller.py", line 199, in _create_trainers_and_managers
self._create_trainer_and_manager(env_manager, behavior_id)
File "e:\my documents\unity\ml-agents\ml-agents\mlagents\trainers\trainer_controller.py", line 173, in _create_trainer_and_manager
trainer = self.trainer_factory.generate(brain_name)
File "e:\my documents\unity\ml-agents\ml-agents\mlagents\trainers\trainer_util.py", line 53, in generate
self.multi_gpu,
File "e:\my documents\unity\ml-agents\ml-agents\mlagents\trainers\trainer_util.py", line 149, in initialize_trainer
run_id,
File "e:\my documents\unity\ml-agents\ml-agents\mlagents\trainers\sac\trainer.py", line 77, in __init__
self._check_param_keys()
File "e:\my documents\unity\ml-agents\ml-agents\mlagents\trainers\trainer.py", line 73, in _check_param_keys
"brain {2}.".format(k, self.__class__, self.brain_name)
mlagents.trainers.trainer.UnityTrainerException: The hyper-parameter vis_encode_type could not be found for the <class 'mlagents.trainers.sac.trainer.SACTrainer
'> trainer of brain Helicopter.
Screenshots
Environment (please complete the following information):
- Windows 7
- ML-Agents v0.14
- tensorflow 1.12.0
- environment: None