Skip to main content

PyTorch Ignite with W&B

Last updated: January 30, 2020

Try the integration in 2 minutes





It's easy to visualize PyTorch Ignite results with Weights & Biases.

Try running this hosted example notebook →

Features in W&B

  1. Line Plots: Comparing metrics across runs
  2. Parallel Coordinates: Visualizing the relationship between hyperparameters and accuracy on different model versions
  3. Run Comparsion: See a table comparing your models
  4. Parameter Importance: See which hyperparameters affect your model the most
14 selected runs


Visualizations




0.100.150.200.250.300.350.400.450.50momentum20406080100120140160180200220240260batch_size05001,0001,5002,0002,5003,0003,5004,000step0.650.700.750.800.850.900.951.00validation accuracy
Showing up to 10 runs
Showing up to 10 runs
14 selected runs