Implement Softmax Cross-Entropy Forward and Backward Passes and a Training Loop

Implement Softmax Cross-Entropy Forward and Backward Passes and a Training Loop

Company: Waymo

Role: Machine Learning Engineer

Category: Machine Learning

Difficulty: medium

Interview Round: Onsite

Overview: Implement the forward and backward passes of softmax with cross-entropy loss from scratch, then write a full training loop for a linear classifier. Tests numerical stability, the logits gradient, vectorized code, gradient checking and minibatch training.