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

Exercise

Writing a training loop

In scikit-learn, the training loop is wrapped in the .fit() method, while in PyTorch, it's set up manually. While this adds flexibility, it requires a custom implementation.

In this exercise, you'll create a loop to train a model for salary prediction.

The show_results() function is provided to help you visualize some sample predictions.

The package imports provided are: pandas as pd, torch, torch.nn as nn, torch.optim as optim, as well as DataLoader and TensorDataset from torch.utils.data.

The following variables have been created: num_epochs, containing the number of epochs (set to 5); dataloader, containing the dataloader; model, containing the neural network; criterion, containing the loss function, nn.MSELoss(); optimizer, containing the SGD optimizer.

Instructions 1/3

undefined XP
    1
    2
    3
  • Write a for loop that iterates over the dataloader; this should be nested within a for loop that iterates over a range equal to the number of epochs.
  • Set the gradients of the optimizer to zero.