I am a little confused about how to get the details of functions from documents. For Example, I want to understand the various options in DataBlock.dataloaders(). How to get information on arguments of the functions.
You can type
doc(DataBlock.dataloaders) to get brief help, plus a link to the full docs, which is here:
Let us know if you still have any questions.
@jeremy Thanks for the reply and amazing first lecture. I still cannot figure out what the second argument
path in the function
path) refers to. I was quite intrigued by ULMFiT, so jumped to the 10th chapter of the book. In one of the code snippets, the path is given.
Taking a look at the docs page, it seems that the
path argument refers to whatever data you are trying to ingest or import into the dataloader (and by extension into the
It seems that the files inside this path are then passed in and go towards creating a
DataLoader object which itself requires a
dataset to be passed in. In any case, I think the
path is just the reference to the data that you’re importing or using for your training.
@strickvl Thanks for the reply. After changing the value of the path, here is what I have observed. The data used to create datasets and
dataloaders is based on the source argument in
path ). The
path gets associated with
dataloaders. So when I passed the
dataloaders to the learner and then checked the
learner.path- then it gave me the value corresponding to
path. So when I used
learner.save , my model was saved in that location. Please correct me if my interpretation was wrong.
In general, is there a place in the documentation where the arguments are explained like what a particular argument does.
You might find it more useful to look at the source code to see what’s going on. I’m not aware of a separate place, beyond what Jeremy mentioned, to find specific explanations of each parameter.
You did a brilliant job of tracking that down - yes that’s exactly correct!