The tests of the wandb-logger are not as comprehensive as for the rest of the loggers. Currently, they just test if constructing a logger works.

Testing wandb in the same way as the other loggers works:

def test_wandb_logger(tmpdir):
    """Verify that basic functionality of wandb logger works."""
    tutils.reset_seed()

    wandb_dir = os.path.join(tmpdir, "wandb")
    logger = WandbLogger(save_dir=wandb_dir, anonymous=True, offline=True)

    hparams = tutils.get_hparams()
    model = LightningTestModel(hparams)

    trainer_options = dict(
        default_save_path=tmpdir,
        max_epochs=1,
        train_percent_check=0.05,
        logger=logger
    )

    trainer = Trainer(**trainer_options)
    result = trainer.fit(model)

    print('result finished')
    assert result == 1, "Training failed"


def test_wandb_pickle(tmpdir):
    """Verify that pickling trainer with wandb logger works."""
    tutils.reset_seed()

    wandb_dir = str(tmpdir)
    logger = WandbLogger(save_dir=wandb_dir, anonymous=True)
    trainer_options = dict(
        default_save_path=tmpdir,
        max_epochs=1,
        logger=logger
    )

    trainer = Trainer(**trainer_options)
    pkl_bytes = pickle.dumps(trainer)
    trainer2 = pickle.loads(pkl_bytes)
    trainer2.logger.log_metrics({"acc": 1.0})

But the CI tests on github for windows fail. The tests for Ubuntu and Mac work. Seems to be some problem related to file-access that I cannot recreate locally