**Fabric is the fast and lightweight way to scale PyTorch models without boilerplate** ______________________________________________________________________

Website • Docs • Getting started • FAQ • Help • Discord

[![PyPI - Python Version](https://img.shields.io/pypi/pyversions/lightning_fabric)](https://pypi.org/project/lightning_fabric/) [![PyPI Status](https://badge.fury.io/py/lightning_fabric.svg)](https://badge.fury.io/py/lightning_fabric) [![PyPI Status](https://pepy.tech/badge/lightning_fabric)](https://pepy.tech/project/lightning_fabric) [![Conda](https://img.shields.io/conda/v/conda-forge/lightning_fabric?label=conda&color=success)](https://anaconda.org/conda-forge/lightning_fabric)
## Maximum flexibility, minimum code changes With just a few code changes, run any PyTorch model on any distributed hardware, no boilerplate! - Easily switch from running on CPU to GPU (Apple Silicon, CUDA, …), TPU, multi-GPU or even multi-node training - Use state-of-the-art distributed training strategies (DDP, FSDP, DeepSpeed) and mixed precision out of the box - All the device logic boilerplate is handled for you - Designed with multi-billion parameter models in mind - Build your own custom Trainer using Fabric primitives for training checkpointing, logging, and more ```diff + import lightning as L import torch import torch.nn as nn from torch.utils.data import DataLoader, Dataset class PyTorchModel(nn.Module): ... class PyTorchDataset(Dataset): ... + fabric = L.Fabric(accelerator="cuda", devices=8, strategy="ddp") + fabric.launch() - device = "cuda" if torch.cuda.is_available() else "cpu model = PyTorchModel(...) optimizer = torch.optim.SGD(model.parameters()) + model, optimizer = fabric.setup(model, optimizer) dataloader = DataLoader(PyTorchDataset(...), ...) + dataloader = fabric.setup_dataloaders(dataloader) model.train() for epoch in range(num_epochs): for batch in dataloader: input, target = batch - input, target = input.to(device), target.to(device) optimizer.zero_grad() output = model(input) loss = loss_fn(output, target) - loss.backward() + fabric.backward(loss) optimizer.step() lr_scheduler.step() ``` ______________________________________________________________________ # Getting started ## Install Lightning
Prerequisites > TIP: We strongly recommend creating a virtual environment first. > Don’t know what this is? Follow our [beginner guide here](https://lightning.ai/docs/stable/install/installation.html). - Python 3.8.x or later (3.8.x, 3.9.x, 3.10.x, ...)
```bash pip install -U lightning ``` ## Convert your PyTorch to Fabric 1. Create the `Fabric` object at the beginning of your training code. ``` import Lightning as L fabric = L.Fabric() ``` 1. Call `setup()` on each model and optimizer pair and `setup_dataloaders()` on all your data loaders. ``` model, optimizer = fabric.setup(model, optimizer) dataloader = fabric.setup_dataloaders(dataloader) ``` 1. Remove all `.to` and `.cuda` calls -> Fabric will take care of it. ```diff - model.to(device) - batch.to(device) ``` 1. Replace `loss.backward()` by `fabric.backward(loss)`. ```diff - loss.backward() + fabric.backward(loss) ``` 1. Run the script from the terminal with ```bash lightning run model path/to/train.py ``` or use the launch() method in a notebook. Learn more about [launching distributed training](https://lightning.ai/docs/fabric/stable/fundamentals/launch.html). ______________________________________________________________________ # FAQ ## When to use Fabric? - **Minimum code changes**- You want to scale your PyTorch model to use multi-GPU or use advanced strategies like DeepSpeed without having to refactor. You don’t care about structuring your code- you just want to scale it as fast as possible. - **Maxmium control**- Write your own training and/or inference logic down to the individual optimizer calls. You aren’t forced to conform to a standardized epoch-based training loop like the one in Lightning Trainer. You can do flexible iteration based training, meta-learning, cross-validation and other types of optimization algorithms without digging into framework internals. This also makes it super easy to adopt Fabric in existing PyTorch projects to speed-up and scale your models without the compromise on large refactors. Just remember: With great power comes a great responsibility. - **Maxmium flexibility**- You want to have full control over your entire training- in Fabric all features are opt-in, and it provides you with a tool box of primitives so you can build your own Trainer. ## When to use the [Lightning Trainer](https://lightning.ai/docs/pytorch/stable/common/trainer.html)? - You want to have all the engineering boilerplate handled for you - dozens of features like checkpointing, logging and early stopping out of the box. Less hassle, less error prone, easy to try different techniques and features. - You want to have good defaults chosen for you - so you can have a better starting point. - You want your code to be modular, readable and well structured - easy to share between projects and with collaborators. ## Can I use Fabric with my LightningModule or Lightning Callback? Yes :) Fabric works with PyTorch LightningModules and Callbacks, so you can choose how to structure your code and reuse existing models and callbacks as you wish. Read more [here](https://lightning.ai/docs/fabric/stable/fundamentals/code_structure.html). ______________________________________________________________________ # Examples - [GAN](https://github.com/Lightning-AI/lightning/tree/master/examples/fabric/dcgan) - [Meta learning](https://github.com/Lightning-AI/lightning/tree/master/examples/fabric/meta_learning) - [Reinforcement learning](https://github.com/Lightning-AI/lightning/tree/master/examples/fabric/reinforcement_learning) - [K-Fold cross validation](https://github.com/Lightning-AI/lightning/tree/master/examples/fabric/kfold_cv) ______________________________________________________________________ ## Asking for help If you have any questions please: 1. [Read the docs](https://lightning.ai/docs/fabric). 1. [Ask a question in our forum](https://lightning.ai/forums/). 1. [Join our discord community](https://discord.com/invite/tfXFetEZxv).