LucasAlegre / morl-baselines

Multi-Objective Reinforcement Learning algorithms implementations.
https://lucasalegre.github.io/morl-baselines
MIT License
296 stars 47 forks source link

PGMORL: Wandb keeps asking for token, even if None is passed #108

Closed wilhem closed 2 months ago

wilhem commented 2 months ago

Take this example here: https://github.com/LucasAlegre/morl-baselines/blob/main/examples/pgmorl_halfcheetah.py and run it (using the very last version of morl-baselines). Running it in a Colab noteboob, it keeps asking for the Wandb-Token all the time, even when passing None:

algo = PGMORL(
    env_id=env_id,
    num_envs=4,
    pop_size=6,
    warmup_iterations=80,
    evolutionary_iterations=20,
    wandb_entity = None,
    num_weight_candidates=7,
    origin=np.array([0.0, -5.0]),
)
ffelten commented 2 months ago

Hello,

If you don't want wandb, you can specify log=False

wilhem commented 2 months ago

Thanks, I oversaw it. But it could be more intuitive, since "log" can be even a log on the screen and not a log for Wandb. From the documentation, it was not clear to me