Notebook 07 - Lesson 217 to train() function writer object has no attribute "add_scalars"

[AxisMeetsWorld]

Hello @mrdbourke (or if there is anyone else I can ping),
Ok, I am at the above mentioned lesson, and tried what I could to troubleshoot this one. I did some searches that didn't seem relevant, and even copied and pasted in code from the book lesson of this lecture. I noticed in this current for loop that references the train() function that it gets to that portion of the train function where you had set up that writer object. I am getting the above error message, and I copied in what I am seeing below as well. What is very odd considering the way Colab is behaving is when I go up to where we defined the train() function, and enter in the code writer. the auto population shows the "add_scalars" function as an option. Why am I getting this error if that is recognized by the code editor? It is a very confusing error message here, and I tried to look at all of the variants we defined, to make sure they were appropriately in place. Please let me know what could be the cause of this error when you get the chance. Here is what I am seeing:

[mrdbourke]

Hi @AxisMeetsWorld ,

This is strange.

Do you know which notebook you're on?

Looks like the writer object is being interpreted as a 'str' object not a SummaryWriter object.

You could try printing print(type(writer)) to see what the writer is being created as?

It should be something like:

from torch.utils.tensorboard import SummaryWriter

# Create a writer with all default settings
writer = SummaryWriter()

print(type(writer))

Output:

<class 'torch.utils.tensorboard.writer.SummaryWriter'>

The SummaryWriter class should definitely have the add_scalars method: https://pytorch.org/docs/stable/tensorboard.html#torch.utils.tensorboard.writer.SummaryWriter.add_scalars

[AxisMeetsWorld]

Unfortunately @mrdbourke,
That didn't turn over any clues:

😞 Can't seem to figure out what was done differently than you either. Here is the code in that train section:
# Train the target model with target dataloader to track experiments # note we are using train() rather than engine.train() train(model=model, train_dataloader=train_dataloader, test_dataloader=test_dataloader, optimizer=optimizer, loss_fn=loss_fn, epochs=epochs, device=device, writer=create_writer(experiment_name=dataloader_name, model_name=model_name, extra=f"{epochs}_epochs"))
And I also checked that area in section 4 where that code for the "train()" was created and that looked to be exactly as you have it:
def train(model: torch.nn.Module, train_dataloader: torch.utils.data.DataLoader, test_dataloader: torch.utils.data.DataLoader, optimizer: torch.optim.Optimizer, loss_fn: torch.nn.Module, epochs: int, device: torch.device) -> Dict[str, List]:

I am at a loss for where to look.