How is data loaded?

[mavaylon1]

Most of the tutorials online show data being wrapped with a DataLoader. In the "01 Fundamentals" section, the data is stored in a tensor. How is this data iterated and loaded into the model? From my understanding the model takes one number at a time from the train X and y when training since we define only one weight and bias.

[mrdbourke]

Hi @mavaylon1 ,

Good question!

For larger datasets, most of the time you'll be using a PyTorch DataLoader (so data be turned into batches, we talk about these later in the course).

But for smaller datasets (like in section 01. PyTorch Workflow Fundamentals), the dataset is so small we don't need to turn it into batches.

This way we can pass the entire dataset X_train and X_test to the model in one go.

[mavaylon1]

What do you mean it takes the entire dataset? With each forward pass it takes only one datapoint at a time, iterating through the set right? Essentially a batch size of 1.

Or are we feeding the entire set as our input at once and not iterating at all?