Summary: Cursor adds advanced PyTorch experiment features

Why It Matters: This change shows Cursor is making its AI coding tools better for complex machine learning work.

What Changed:

• The "Careers" link changed from an external site to an internal page.
• The interactive demo for the IDE now shows "In Progress 3" and "Ready for Review 3" instead of "In Progress 4" and "Ready for Review 2".
• The PyTorch MNIST Experiments section was greatly expanded. It now includes:
  • More detailed code for train_model.py with advanced features like mixed precision training, learning rate scheduling, validation splitting, gradient clipping, and checkpointing.
  • Mentions of updating evaluation.py and creating run_experiment.py.
  • A summary of new features like YAML config, saved history, confusion matrix, and classification reports.
• The "Access the best models" section now lists "Composer 1" as an auto-suggested model instead of "GPT-5 High".
• The "Complete codebase understanding" section added "Choose a model" and "Searched Where is the model picker UI implemented?" to its demo text.
• The "Join us" link changed from an external site to an internal careers page.