Built to make you extraordinarily productive,
Cursor is the best way to code with AI.

hero Image
                  ():
  parser = argparse.ArgumentParser(description='Run MNIST experiment')
  parser.add_argument('--config', type=str, default='experiments/config.yaml',
                      help='Path to config file')
  parser.add_argument('--device', type=str, default=None,
                      help='Device to use (cuda/cpu)')
  args = parser.parse_args()
  
  print("Starting training...")
  model, history, test_loader = train_model(args.config)
  
  print("\nEvaluating model...")
  device = model.device if hasattr(model, 'device') else next(model.parameters()).device
  accuracy, cm = evaluate_model(model, test_loader, device)
  
  print(f"\nFinal test accuracy: {accuracy:.4f}")
  print("Results saved to ./results/")
 
                
Cursor
Get Cursor

In Progress 3

Loader SVG

Enterprise Order Management System

Generating

Loader SVG

Analyze Tab vs Agent Usage Patterns

Generating

Loader SVG

Fix PR Comments Fetching Issue

Generating

Ready For Preview 3

Circle Tick SVG

PyTorch MNIST Experiments

+162 -37 . Done, configurable MNIST experiment framework with AMP and reports. • **Training**: AMP, train/val split, cosine schedu

now
circle-tick SVG

Set up Cursor Rules for Dashboard

+37 -0 . Set up Cursor Rules for Dashboard

30m
Loader SVG

Bioinformatics Tools

+135 -21 . Biometrics Tools

45m

PyTorch MNIST Experiments

Add mixed precision training, learning rate scheduling, and proper validation. Also create an experiment config system so I can easily run different hyperparameter settings.

Thought

10s

Searched

PyTorch mixed precision training best practices

I'll enhance your MNIST trainer with a complete experiment framework including mixed precision, validation splitting, and proper configuration management. Let me start by rewriting the training module:

File SVG

train_model.py

+156 -34

Now let me update the evaluation module to save results and generate a detailed report:

File SVG

evaluation.py

+24 -6

Let me create a command-line runner script to make experiments easy to execute:

File SVG

run_experiment.py

+6

Done, configurable MNIST experiment framework with AMP and reports.

  • Training : AMP, train/val split, cosine schedule, gradient clipping, checkpoints
  • Experimentation : YAML config, saved history, confusion matrix + classification report, CLI runner

                  
                    #!/usr/bin/env python3
import argparse
from train_model import train_model
from evaluation import evaluate_model

def main():
  parser = argparse.ArgumentParser(description='Run MNIST experiment')
  parser.add_argument('--config', type=str, default='experiments/config.yaml',
                      help='Path to config file')
  parser.add_argument('--device', type=str, default=None,
                      help='Device to use (cuda/cpu)')
  args = parser.parse_args()
  
  print("Starting training...")
  model, history, test_loader = train_model(args.config)
  
  print("\nEvaluating model...")
  device = model.device if hasattr(model, 'device') else next(model.parameters()).device
  accuracy, cm = evaluate_model(model, test_loader, device)
  
  print(f"\nFinal test accuracy: {accuracy:.4f}")
  print("Results saved to ./results/")

if __name__ == "__main__":
  main()
                

Trusted every day by millions of professional developers.

  • Stripe Logo
  • Open ai Logo
  • Linear Logo
  • Datadog Logo
  • Nvidia Logo
  • Figma Logo
  • Ramp Logo
  • Adobe Logo
Feature1
Feature2

Everywhere software gets built

Cursor is in GitHub reviewing your PRs, a teammate in Slack, and anywhere else you work.

Learn about Cursor's ecosystem
Feature3

The new way to build software.

It was night and day from one batch to another, adoption went from single digits to over 80%. It just spread like wildfire, all the best builders were using Cursor.

diana-hu-avatar

Diana Hu

General Partner, Y Combinator

The most useful AI tool that I currently pay for, hands down, is Cursor. It's fast, autocompletes when and where you need it to, handles brackets properly, sensible keyboard shortcuts, bring-your-own-model... everything is well put together.

shadcn-avatar

shadcn

Creator of shadcn/ui

The best LLM applications have an autonomy slider: you control how much independence to give the AI. In Cursor, you can do Tab completion, Cmd+K for targeted edits, or you can let it rip with the full autonomy agentic version.

andrej-karpathy-avatar

Andrej Karpathy

CEO, Eureka labs

Cursor quickly grew from hundreds to thousands of extremely enthusiastic Stripe employees. We spend more on R&D and software creation than any other undertaking, and there's significant economic outcomes when making that process more efficient and productive.

patrick-collison-avatar

Patrick Collison

CEO, Stripe

It's official.

I hate vibe coding.

I love Cursor tab coding.

It's wild.

theprimeagen-avatar

ThePrimeagen

@theprimeagen

It's definitely becoming more fun to be a programmer. It's less about digging through pages and more about what you want to happen. We are at the 1% of what's possible, and it's in interactive experiences like Cursor where models like GPT-5 shine brightest.

greg-brockman-avatar

Greg Brockman

President, OpenAI

Stay on the frontier

Access the best models

Choose between every cutting-edge model from OpenAI, Anthropic, Gemini, and xAI.

Explore models
use case 1 png

Complete codebase understanding

Cursor learns how your codebase works, no matter the scale or complexity.

Learn about semantic search
use case 2 png

Develop enduring software

Trusted by over half of the Fortune 500 to accelerate development, securely and at scale.

Explore enterprise
use case 2 png

Changelog

Subagents, Skills, and Image Generation

CLI Agent Modes and Cloud Handoff

New CLI Features and Improved CLI Performance

Layout Customization and Stability Improvements

See what's new in Cursor

Cursor is an applied team focused
on building the future of coding.

Home Page showing the team photo

Recent highlights

Introducing Cursor 2.0 and Composer

A new interface and our first coding model, both purpose-built for working with agents.

Product .

Improving Cursor Tab with online RL

Our new Tab model makes 21% fewer suggestions while having 28% higher accept rate.

Research .

1.5x faster MoE training with custom MXFP8 kernels

Achieving a 3.5x MoE layer speedup with a complete rebuild for Blackwell GPUs.

Research .

View more posts

Try Cursor Now.