def get_variant_spec(args):
algorithm = args.algorithm
layer_size = 128
variant_spec = {
'layer_size':
layer_size,
'policy_params': {
'type': 'GaussianPolicy',
'kwargs': {
'hidden_layer_sizes': (layer_size, layer_size),
'squash': True,
},
},
'algorithm_params':
deep_update(ALGORITHM_PARAMS_BASE,
ALGORITHM_PARAMS_ADDITIONAL.get(algorithm, {})),
'Q_params': {
'type': 'double_feedforward_Q_function',
'kwargs': {
'hidden_layer_sizes': (layer_size, layer_size),
},
},
'run_params': {
'seed': 1,
},
}
return variant_spec
def get_variant_spec_base(universe, domain, task, policy, algorithm):
algorithm_params = deep_update(
ALGORITHM_PARAMS_BASE,
ALGORITHM_PARAMS_ADDITIONAL.get(algorithm, {}),
get_algorithm_params(universe, domain, task),
)
variant_spec = {
'git_sha': get_git_rev(__file__),
'environment_params': {
'training': {
'domain': domain,
'task': task,
'universe': universe,
'kwargs': get_environment_params(universe, domain, task),
},
'evaluation': tune.sample_from(lambda spec: (
spec.get('config', spec)
['environment_params']
['training']
)),
},
'policy_params': tune.sample_from(get_policy_params),
'exploration_policy_params': {
'type': 'UniformPolicy',
'kwargs': {
'observation_keys': tune.sample_from(lambda spec: (
spec.get('config', spec)
['policy_params']
['kwargs']
.get('observation_keys')
))
},
},
'Q_params': {
'type': 'double_feedforward_Q_function',
'kwargs': {
'hidden_layer_sizes': (M, M),
'observation_keys': None,
'observation_preprocessors_params': {}
},
},
'algorithm_params': algorithm_params,
'replay_pool_params': {
'type': 'SimpleReplayPool',
'kwargs': {
'max_size': int(1e6),
},
},
'sampler_params': {
'type': 'SimpleSampler',
'kwargs': {
'max_path_length': get_max_path_length(universe, domain, task),
}
},
'run_params': {
'host_name': get_host_name(),
'seed': tune.sample_from(
lambda spec: np.random.randint(0, 10000)),
'checkpoint_at_end': True,
'checkpoint_frequency': tune.sample_from(get_checkpoint_frequency),
'checkpoint_replay_pool': False,
},
}
return variant_spec
def get_variant_spec_base(env, randomized, use_predictive_model,
observation_mode, reward_type, single_obj_reward,
all_random, trimodal_positions_choice, num_objects,
model_dir, num_execution_per_step, policy,
algorithm):
algorithm_params = deep_update(
ALGORITHM_PARAMS_BASE,
ALGORITHM_PARAMS_ADDITIONAL.get(algorithm, {}),
get_algorithm_params_roboverse(env, use_predictive_model),
)
variant_spec = {
'git_sha': get_git_rev(__file__),
'environment_params': {
'training': {
'env': env,
'randomize_env': randomized,
'use_predictive_model': use_predictive_model,
'obs': observation_mode,
'reward_type': reward_type,
'single_obj_reward': single_obj_reward,
'all_random': all_random,
'trimodal_positions_choice': trimodal_positions_choice,
'num_objects': num_objects,
'model_dir': model_dir,
'num_execution_per_step': num_execution_per_step,
'kwargs': {
'image_shape': (48, 48, 3)
}
},
'evaluation':
tune.sample_from(lambda spec: (spec.get('config', spec)[
'environment_params']['training'])),
},
# 'policy_params': tune.sample_from(get_policy_params),
'policy_params': {
'class_name': 'FeedforwardGaussianPolicy',
'config': {
'hidden_layer_sizes': (M, M),
'squash': False, #True,
'observation_keys': None,
'preprocessors': None,
},
},
'exploration_policy_params': {
'class_name': 'ContinuousUniformPolicy',
'config': {
'observation_keys':
tune.sample_from(lambda spec: (spec.get('config', spec)[
'policy_params']['config'].get('observation_keys')))
},
},
'Q_params': {
'class_name': 'double_feedforward_Q_function',
'config': {
'hidden_layer_sizes': (M, M),
'observation_keys': None,
'preprocessors': None,
},
},
'algorithm_params': algorithm_params,
'replay_pool_params': {
'class_name': 'SimpleReplayPool',
'config': {
'max_size': int(1e6),
},
},
'sampler_params': {
'class_name': 'SimpleSampler',
'config': {
'max_path_length':
get_max_path_length_roboverse(env, use_predictive_model),
}
},
'run_params': {
'host_name': get_host_name(),
'seed': tune.sample_from(lambda spec: np.random.randint(0, 10000)),
'checkpoint_at_end': True,
'checkpoint_frequency': tune.sample_from(get_checkpoint_frequency),
'checkpoint_replay_pool': False,
},
}
return variant_spec
def get_variant_spec_base(universe, domain, task, policy, algorithm):
algorithm_params = deep_update(
ALGORITHM_PARAMS_BASE,
ALGORITHM_PARAMS_ADDITIONAL.get(algorithm, {}),
get_algorithm_params(universe, domain, task),
)
variant_spec = {
# doodad is complaining about this so we're just gonna hardcode a SHA
#'git_sha': 'bd1dc29e166aca501af2e58a5057418126a3e435 master', #get_git_rev(__file__),
'environment_params': {
'training': {
'domain': domain,
'task': task,
'universe': universe,
'kwargs': get_environment_params(universe, domain, task),
},
'evaluation':
tune.sample_from(lambda spec: (spec.get('config', spec)[
'environment_params']['training'])),
},
# 'policy_params': tune.sample_from(get_policy_params),
'policy_params': {
'class_name': 'FeedforwardGaussianPolicy',
'config': {
'hidden_layer_sizes': (M, M),
'squash': True,
'observation_keys': None,
'preprocessors': None,
},
},
'exploration_policy_params': {
'class_name': 'ContinuousUniformPolicy',
'config': {
'observation_keys':
tune.sample_from(lambda spec: (spec.get('config', spec)[
'policy_params']['config'].get('observation_keys')))
},
},
'Q_params': {
'class_name': 'double_feedforward_Q_function',
'config': {
'hidden_layer_sizes': (M, M),
'observation_keys': None,
'preprocessors': None,
},
},
'algorithm_params': algorithm_params,
'replay_pool_params': {
'class_name': 'SimpleReplayPool',
'config': {
'max_size': int(1e6),
},
},
'sampler_params': {
'class_name': 'SimpleSampler',
'config': {
'max_path_length': get_max_path_length(universe, domain, task),
}
},
'run_params': {
'host_name': get_host_name(),
'seed': tune.sample_from(lambda spec: np.random.randint(0, 10000)),
'checkpoint_at_end': True,
'checkpoint_frequency': tune.sample_from(get_checkpoint_frequency),
'checkpoint_replay_pool': False,
},
}
return variant_spec
