How-to Guides

Avoid overfitting

Learn how to add validation and test loops

Build a model

Step by step guide to build your model

Configure hyperparameters from the CLI

Make your experiments modular via command line interface

Customize the progress bar

Change the progress bar monitoring and tracking

Deploy models into production

Deploy models with different levels of scal

Optimize training

Explore advanced training techniques

Find bottlenecks in your code

Learn how to profile your experiments to find bottlenecks

Finetune a model

Learn how to use pretrained models

Manage data

How to use basic to advanced data techniques

Manage experiments

Learn to track and visualize with experiment managers

Organize existing PyTorch into Lightning

Convert your vanila PyTorch to Lightning

Run on a multi-node cluster

Learn to run on multi-node in the cloud or on your cluster

Save and load model progress

Save and load progress with checkpoints

Save memory with half-precision

Use precision techniques to train faster and save memory

Set up large models efficiently

Avoid memory peaks and speed up the initialization of large models

Train models with billions of parameters

Scale GPU training for models with billions of parameters

Train in a notebook

Train models in interactive notebooks (Jupyter, Colab, Kaggle, etc.)

Train on single or multiple GPUs

Train models faster with GPU accelerators

Train on single or multiple HPUs

Train models faster with HPU accelerators

Train on single or multiple TPUs

TTrain models faster with TPU accelerators

Train on MPS

Train models faster with Apple Silicon GPUs

Track and Visualize Experiments

Learn to track and visualize experiments

Use a pretrained model

Improve results with transfer learning on pretrained models

Use a pure PyTorch training loop

Run your pure PyTorch loop with Lightning

Truncated Back-Propagation Through Time

Efficiently step through time when training recurrent models