1. Learn
    2. /
  2. Courses
    3. /
  3. Introduction to Deep Learning with PyTorch

Exercise

From data loading to running a forward pass

In this exercise, you'll create a PyTorch DataLoader from a pandas DataFrame and call a model on this dataset. Specifically, you'll run a forward pass on a neural network. You'll continue working with fully connected neural networks, as you have done so far.

You'll begin by subsetting a loaded DataFrame called dataframe, converting features and targets NumPy arrays, and converting to PyTorch tensors in order to create a PyTorch dataset.

This dataset can be loaded into a PyTorch DataLoader, batched, shuffled, and used to run a forward pass on a custom fully connected neural network.

NumPy as np, pandas as pd, torch, TensorDataset(), and DataLoader() have been imported for you.

Instructions 1/3

undefined XP
    1
    2
    3
  • Extract the features (ph, Sulfate, Conductivity, Organic_carbon) and target (Potability) values and load them into the appropriate tensors to represent features and targets.
  • Use both tensors to create a PyTorch dataset using the dataset class that's quickest to use when tensors don't require any additional preprocessing.