harl.common package

Subpackages

Submodules

harl.common.base_logger module

Base logger.

class harl.common.base_logger.BaseLogger(args, algo_args, env_args, num_agents, writter, run_dir)[source]

Bases: object

Base logger class. Used for logging information in the on-policy training pipeline.

close()[source]

Close the logger.

episode_init(episode)[source]

Initialize the logger for each episode.

episode_log(actor_train_infos, critic_train_info, actor_buffer, critic_buffer)[source]

Log information for each episode.

eval_init()[source]

Initialize the logger for evaluation.

eval_init_off_policy(total_num_steps)[source]
eval_log(eval_episode)[source]

Log evaluation information.

eval_per_step(eval_data)[source]

Log evaluation information per step.

eval_thread_done(tid)[source]

Log evaluation information.

get_task_name()[source]

Get the task name.

init(episodes)[source]

Initialize the logger.

log_env(env_infos)[source]

Log environment information.

log_train(actor_train_infos, critic_train_info)[source]

Log training information.

per_step(data)[source]

Process data per step.

harl.common.valuenorm module

ValueNorm.

class harl.common.valuenorm.ValueNorm(input_shape, norm_axes=1, beta=0.99999, per_element_update=False, epsilon=1e-05, device=device(type='cpu'))[source]

Bases: Module

Normalize a vector of observations - across the first norm_axes dimensions

denormalize(input_vector)[source]

Transform normalized data back into original distribution

normalize(input_vector)[source]
running_mean_var()[source]

Get running mean and variance.

update(input_vector)[source]

Module contents