Language Language English 简体中文

unilab.logging.offpolicy.OffPolicyLogger

class unilab.logging.offpolicy.OffPolicyLogger

Bases: BaseTrainingLogger

Rich logger for off-policy RL algorithms (SAC, TD3, etc).

Parameters:

• algo_name (str)

• max_iterations (int)

• num_envs (int)

• env_name (str)

• obs_dim (int)

• action_dim (int)

• refresh_per_second (int)

• log_dir (str)

• log_backend (str)

• wandb_project (str)

• wandb_entity (str | None)

• wandb_name (str)

• wandb_group (str | None)

• wandb_job_type (str | None)

• wandb_tags (list[str] | None)

• wandb_notes (str | None)

Methods

__init__([algo_name, max_iterations, ...])
close()	Release live terminal state and backend handles without printing a summary.
finish(*[, title, extra_summary])
log_buffer_fill(current, target)
log_collector(total_steps, buffer_size[, ...])
log_save(path)
log_status(status)
log_step(iteration[, metrics, reward, ...])
set_collection_sync(enabled[, ...])
start(*[, status])
update_buffer_utilization(utilization)
update_collector_timing(timing_ms)
update_done_rates(timeout_rate, terminated_rate)
update_ep_length(length)
update_replay_queue(current_len, max_size)
update_staging_pool(current_len, max_size)

__init__(algo_name='RL', max_iterations=1500, num_envs=4096, env_name='', obs_dim=0, action_dim=0, refresh_per_second=4, log_dir='', log_backend='tensorboard', wandb_project='unilab', wandb_entity=None, wandb_name='', wandb_group=None, wandb_job_type=None, wandb_tags=None, wandb_notes=None)

Parameters:

• algo_name (str)

• max_iterations (int)

• num_envs (int)

• env_name (str)

• obs_dim (int)

• action_dim (int)

• refresh_per_second (int)

• log_dir (str)

• log_backend (str)

• wandb_project (str)

• wandb_entity (str | None)

• wandb_name (str)

• wandb_group (str | None)

• wandb_job_type (str | None)

• wandb_tags (list[str] | None)

• wandb_notes (str | None)

start(*, status='Warming up...')

Parameters:

status (str)

finish(*, title='Training Summary', extra_summary='')

Parameters:

• title (str)

• extra_summary (str)

log_buffer_fill(current, target)

Parameters:

• current (int)

• target (int)

update_collector_timing(timing_ms)

Parameters:

timing_ms (dict[str, float])

update_done_rates(timeout_rate, terminated_rate)

Parameters:

• timeout_rate (float)

• terminated_rate (float)

update_buffer_utilization(utilization)

Parameters:

utilization (float)

update_replay_queue(current_len, max_size)

Parameters:

• current_len (int)

• max_size (int)

update_staging_pool(current_len, max_size)

Parameters:

• current_len (int)

• max_size (int)

set_collection_sync(enabled, env_steps_per_sync=0)

Parameters:

• enabled (bool)

• env_steps_per_sync (int)

log_collector(total_steps, buffer_size, mean_reward=0.0)

Parameters:

• total_steps (int)

• buffer_size (int)

• mean_reward (float)

log_step(iteration, metrics=None, reward=None, reward_metrics=None, reward_components=None, train_time=0.0, wait_time=0.0, learner_incremental_h2d_time=0.0, weight_sync_time=0.0, extra_info=None)

Parameters:

• iteration (int)

• metrics (dict[str, float] | None)

• reward (float | None)

• reward_metrics (dict[str, float] | None)

• reward_components (dict[str, float] | None)

• train_time (float)

• wait_time (float)

• learner_incremental_h2d_time (float)

• weight_sync_time (float)

• extra_info (dict | None)

log_status(status)

Parameters:

status (str)

close()

Release live terminal state and backend handles without printing a summary.

Return type:

None

log_save(path)

Parameters:

path (str)

update_ep_length(length)

Parameters:

length (float)