Summary: Cursor adds advanced PyTorch experiment framework

Why It Matters: This change shows Cursor is improving its AI coding tools for complex machine learning tasks.

What Changed:

• The PyTorch MNIST example code was greatly expanded.
• The new code includes mixed precision training, learning rate scheduling, and proper validation.
• It adds an experiment config system for hyperparameter settings.
• The code now supports data augmentation and dropout in the model.
• It includes saving checkpoints and training history.
• The description of the AI agent's actions for the PyTorch example was updated to reflect these new features.
• The "Complete codebase understanding" section added new example text: "Choose a model * Searched Where is the model picker UI implemented? * Searching".
• The changelog entry for "1.7 Sep 29, 2025" changed from "Agent Autocomplete, Hooks, and Team Rules" to "Browser Controls, Agent Autocomplete, and Hooks".