wandb

W&B utilities for Neuroevolution fitting.

cneuromax.fitting.neuroevolution.utils.wandb.setup_wandb(logger, output_dir)

Sets up W&B logging for all MPI processes.

Parameters:

• logger (Callable[..., Any]) – See wandb.init().

• output_dir (str) – See output_dir.

Return type:

None

cneuromax.fitting.neuroevolution.utils.wandb.terminate_wandb()

Terminates W&B logging.

Return type:

None

cneuromax.fitting.neuroevolution.utils.wandb.gather(logged_score, curr_gen, agent_total_num_steps)

Gathers logged scores from all MPI processes.

Parameters:

• logged_score (float | None) – A value logged during evaluation. If None, then no value was logged during evaluation.

• curr_gen (int) – See curr_gen.

• agent_total_num_steps (int) – See total_num_steps.

Return type:

None