commonpower.control.logging_utils.callbacks.MARLBaseCallback

class MARLBaseCallback(verbose: int = 0)

Bases: object

Base class for a multi-agent callback. Adapted from stable-baselines3 BaseCallback https://github.com/DLR-RM/stable-baselines3/blob/master/stable_baselines3/common/callbacks.py

Parameters:

verbose (int) – Verbosity level: 0 for no output, 1 for info messages, 2 for debug messages

Methods

init_callback	Initialize the callback by saving references to the RL runner and the training environment for convenience.
on_rollout_end	Any operations the callback has to perform at the end of one training episode
on_rollout_start
on_step	This method will be called by the runner after each call to env.step().
on_training_end	Any operations the callback has to perform after the training is finished
on_training_start	Any operations the callback has to perform before the training starts
update_child_locals	Update the references to the local variables on sub callbacks.
update_locals	Update the references to the local variables.
update_num_timesteps

Attributes

runner
logger

_on_rollout_end() → None

At the end of one training episode, we want to log some information about the safety shield.

Returns:

None

_on_step() → bool

Internal operation that should be performed in each step

Returns:

(bool) – If the callback returns False, training is aborted early.

init_callback(runner: runners.BaseRunner) → None

Initialize the callback by saving references to the RL runner and the training environment for convenience.

on_rollout_end() → None

Any operations the callback has to perform at the end of one training episode

Returns:

None

on_step(num_timesteps: int) → bool

This method will be called by the runner after each call to env.step().

Parameters:

num_timesteps (int) – Number of environments * number of steps per env

Returns:

If the callback returns False, training is aborted early.

on_training_end() → None

Any operations the callback has to perform after the training is finished

Returns:

None

on_training_start(locals_: Dict[str, Any], globals_: Dict[str, Any], num_timesteps: int = 0) → None

Any operations the callback has to perform before the training starts

Parameters:

• locals (Dict[str, Any]) – local variables

• globals (Dict[str, Any]) – global variables

• num_timesteps (int) – current training progress

Returns:

None

update_child_locals(locals_: Dict[str, Any]) → None

Update the references to the local variables on sub callbacks.

Parameters:

• (Dict[str – the local variables during rollout collection

• Any]) – the local variables during rollout collection

update_locals(locals_: Dict[str, Any]) → None

Update the references to the local variables.

Parameters:

• (Dict[str – the local variables during rollout collection

• Any]) – the local variables during rollout collection

update_num_timesteps(num_timesteps: int = 0) → None

Parameters:

num_timesteps (int) – number of environments * number of time steps (training progress)

Returns:

None