Tensorboard: PyTorch writer.add_scalars() does not work as intented.

I tried the code as follows:

from torch.utils.tensorboard import SummaryWriter
import wandb
wandb.tensorboard.patch(tensorboardX=False, pytorch=True)
wandb.init()
writer = SummaryWriter()
scalars = {
    'scalar_1': 1.0,
    'scalar_2': 2.0,
    'scalar_3': 3.0
}
writer.add_scalars('test', scalars, global_step=1)
writer.close()
wandb.finish()

My observations:

1. There is a warning that I face with my implementation which is as follows:

wandb: WARNING When using several event log directories, please call wandb.tensorboard.patch(root_logdir="...") before wandb.init

I have tried figuring out the root_dir parameter, due to the unavailability of a docstring or documentation, I was not able to remove the warning. 2. If we ignore the warning the tensorboard integration in wandb does not really work like the one in colab.

The file structure of the event files that are logged by the SummaryWriter are	But in the wandb folder the event files are not structured the same way

Here we do not see the folders namely test_scalar_1, test_scalar_2, test_scalar_3.

This leads us to the bug in the tensorboard integration

Colab tensorboard	Tensorboard in wandb

Here we do not notice the anything in the runs section of wandb tensorboard.

I am suspecting a probable issue with https://github.com/wandb/client/blob/30f77c8a320fe42a519f76eb012e61101afad9ba/wandb/integration/tensorboard/monkeypatch.py#L90

[issue-label-bot[bot]]

Issue-Label Bot is automatically applying the label bug to this issue, with a confidence of 0.88. Please mark this comment with 👍 or 👎 to give our bot feedback!

Links: app homepage, dashboard and code for this bot.

[raubitsj]

This fix will go out in the 0.10.13 release