lightning/docs/source/common_usecases.rst

166 lines
5.6 KiB
ReStructuredText

:orphan:
Common use cases
================
.. include:: links.rst
.. raw:: html
<div class="tutorials-callout-container">
<div class="row">
.. Add callout items below this line
.. customcalloutitem::
:description: Learn to train Lightning models on the cloud
:header: Cloud Training
:button_link: clouds/cloud_training.html
:card_style: text-container-small
.. customcalloutitem::
:description: Lightning checkpoints have everything you need to save and restore your models
:header: Checkpointing
:button_link: common/checkpointing.html
:card_style: text-container-small
.. customcalloutitem::
:description: Learn to train on your university or company's cluster
:header: Cluster Training
:button_link: clouds/cluster.html
:card_style: text-container-small
.. customcalloutitem::
:description: Tricks for debugging your Lightning Models
:header: Debugging
:button_link: common/debugging.html
:card_style: text-container-small
.. customcalloutitem::
:description: Save time and money by training until key metrics stop improving or time has elapsed
:header: Early Stopping
:button_link: common/early_stopping.html
:card_style: text-container-small
.. customcalloutitem::
:description: Here you'll find the latest SOTA training techniques such as SWA, accumulated gradients, etc...
:header: Effective Training Techniques
:button_link: advanced/training_tricks.html
:card_style: text-container-small
.. customcalloutitem::
:description: Avoid over-fitting (memorizing the dataset) with these techniques
:header: Evaluation
:button_link: common/evaluation.html
:card_style: text-container-small
.. customcalloutitem::
:description: Before coding a complex model, use lightning-flash to create a baseline in a few lines of code
:header: Fast Baselines
:button_link: ecosystem/flash.html
:card_style: text-container-small
.. customcalloutitem::
:description: Enable fault-tolerant training in clusters/clouds where machines might fail (ie: pre-emtible machines)
:header: Fault-Tolerant Training
:button_link: advanced/fault_tolerant_training.html
:card_style: text-container-small
.. customcalloutitem::
:description: Make your models more flexible by enabling command-line arguments
:header: Hyperparameters (via command-line)
:button_link: common/hyperparameters.html
:card_style: text-container-small
.. customcalloutitem::
:description: Use the latest tricks to easily productionize your Lightning models
:header: Inference in Production
:button_link: common/production_inference.html
:card_style: text-container-small
.. customcalloutitem::
:description: Reduce configuration boilerplate with the Lightning CLI
:header: Lightning CLI
:button_link: common/lightning_cli.html
:card_style: text-container-small
.. customcalloutitem::
:description: Visualize your machine learning experiments with these experiment managers
:header: Loggers (experiment managers)
:button_link: common/loggers.html
:card_style: text-container-small
.. customcalloutitem::
:description: Use the model registry to mix and match your models and Datamodules
:header: Model and Datamodule Registry
:button_link: common/lightning_cli.html#multiple-models-and-or-datasets
:card_style: text-container-small
.. customcalloutitem::
:description: Train 1TB+ parameter models with these advanced built-in techniques
:header: Model Parallelism
:button_link: advanced/model_parallel.html
:card_style: text-container-small
.. customcalloutitem::
:description: Increase batch-sizes and improve speeds by training using 16-bit precision and more
:header: N-Bit Precision
:button_link: advanced/precision.html
:card_style: text-container-small
.. customcalloutitem::
:description: Enable manual optimization to fully control the optimization procedure for advanced research
:header: Manual Optimization
:button_link: common/optimization.html
:card_style: text-container-small
.. customcalloutitem::
:description: Use these profilers to find bottlenecks in your model
:header: Profiling
:button_link: advanced/profiler.html
:card_style: text-container-small
.. customcalloutitem::
:description: Use these built-in progress bars or learn how to make your own!
:header: Progress Bar
:button_link: common/progress_bar.html
:card_style: text-container-small
.. customcalloutitem::
:description: Compress model sizes to speed up model inference for deployment without loss of performance (accuracy)
:header: Pruning and Quantization
:button_link: advanced/pruning_quantization.html
:card_style: text-container-small
.. customcalloutitem::
:description: Work with data on any local or cloud filesystem such as S3 on AWS, GCS on Google Cloud, or ADL on Azure
:header: Remote Filesystems
:button_link: common/remote_fs.html
:card_style: text-container-small
.. customcalloutitem::
:description: Building the next Deepspeed, FSDP or fancy scaling technique? Add them to Lightning here
:header: Strategy Registry
:button_link: advanced/strategy_registry.html
:card_style: text-container-small
.. customcalloutitem::
:description: Simplify metrics calculations to scale-proof your models
:header: Torchmetrics
:button_link: ecosystem/metrics.html
:card_style: text-container-small
.. customcalloutitem::
:description: Use models training on large datasets to achieve better results when you don't have much data
:header: Transfer Learning (finetuning)
:button_link: advanced/transfer_learning.html
:card_style: text-container-small
.. raw:: html
</div>
</div>
.. End of callout item section