Understand it.
Build it.
Make it work.

Choose what you want to learn

Try a PyTorch problem

Read the challenge. Make it work.

Start with the provided code, work through the requirements, and check your solution against real test cases on your machine.

import torch

def accuracy(predictions, labels):
    """Multiclass classification accuracy.

    predictions: shape (N, C): class scores (logits or probabilities).
    labels:      shape (N,): int class indices (delivered as floats by the helper).

    Returns: scalar float: fraction of predictions whose argmax equals the label.
    """
    pass

By category

Learn with others

Choose a track with friends, compare approaches, and help each other get unstuck. Share your progress on a squad board and keep the conversation in one place.