Common Utilities

Common Utilities in RLzoo

Functions for utilization.

# Requirements tensorflow==2.0.0a0 tensorlayer==2.0.1

rlzoo.common.utils.call_default_params(env, envtype, alg, default_seed=True)

Get the default parameters for training from the default script

rlzoo.common.utils.get_algorithm_module(algorithm, submodule)

Get algorithm module in the corresponding folder

rlzoo.common.utils.load_model(model, model_name, algorithm_name, env_name)

load saved neural network model

Parameters:

• model – tensorlayer.models.Model
• model_name – string, e.g. ‘model_sac_q1’
• algorithm_name – string, e.g. ‘SAC’

rlzoo.common.utils.make_env(env_id)

rlzoo.common.utils.parse_all_args(parser)

Parse known and unknown args

rlzoo.common.utils.plot(episode_rewards, algorithm_name, env_name)

plot the learning curve, saved as ./img/algorithm_name-env_name.png

Parameters:

• episode_rewards – array of floats
• algorithm_name – string
• env_name – string

rlzoo.common.utils.plot_save_log(episode_rewards, algorithm_name, env_name)

plot the learning curve, saved as ./img/algorithm_name-env_name.png, and save the rewards log as ./log/algorithm_name-env_name.npy

Parameters:

• episode_rewards – array of floats
• algorithm_name – string
• env_name – string

rlzoo.common.utils.save_model(model, model_name, algorithm_name, env_name)

save trained neural network model

Parameters:

• model – tensorlayer.models.Model
• model_name – string, e.g. ‘model_sac_q1’
• algorithm_name – string, e.g. ‘SAC’

rlzoo.common.utils.set_seed(seed, env=None)

set random seed for reproduciblity