lightning

The Deep Learning framework to train, deploy, and ship AI products Lightning fast.

Apache-2.0 171 个版本 Python >=3.10

Lightning AI et al. <developer@lightning.ai>

安装

pip install lightning

poetry add lightning

pipenv install lightning

conda install lightning

描述

The deep learning framework to pretrain and finetune AI models.

Serving models? Use LitServe to build custom inference servers in pure Python.

Quick start • Examples • PyTorch Lightning • Fabric • Lightning Cloud • Community • Docs

GitHub commit activity

Why PyTorch Lightning?

Training models in plain PyTorch requires writing and maintaining a lot of repetitive engineering code. Handling backpropagation, mixed precision, multi-GPU, and distributed training is error-prone and often reimplemented for every project. PyTorch Lightning organizes PyTorch code to automate this infrastructure while keeping full control over your model logic. You write the science. Lightning handles the engineering, and scales from CPU to multi-node GPUs without changing your core code. PyTorch experts can still opt into expert-level control.

Fun analogy: If PyTorch is Javascript, PyTorch Lightning is ReactJS or NextJS.

Looking for GPUs?

Lightning Cloud is the easiest way to run PyTorch Lightning without managing infrastructure. Start training with one command and get GPUs, autoscaling, monitoring, and a free tier. No cloud setup required.

You can also run PyTorch Lightning on your own hardware or cloud.

Lightning has 2 core packages

PyTorch Lightning: Train and deploy PyTorch at scale.
Lightning Fabric: Expert control.

Lightning gives you granular control over how much abstraction you want to add over PyTorch.

Quick start

Install Lightning:

pip install lightning

PyTorch Lightning example

Define the training workflow. Here's a toy example (explore real examples):

# main.py
# ! pip install torchvision
import torch, torch.nn as nn, torch.utils.data as data, torchvision as tv, torch.nn.functional as F
import lightning as L

# --------------------------------
# Step 1: Define a LightningModule
# --------------------------------
# A LightningModule (nn.Module subclass) defines a full *system*
# (ie: an LLM, diffusion model, autoencoder, or simple image classifier).


class LitAutoEncoder(L.LightningModule):
    def __init__(self):
        super().__init__()
        self.encoder = nn.Sequential(nn.Linear(28 * 28, 128), nn.ReLU(), nn.Linear(128, 3))
        self.decoder = nn.Sequential(nn.Linear(3, 128), nn.ReLU(), nn.Linear(128, 28 * 28))

    def forward(self, x):
        # in lightning, forward defines the prediction/inference actions
        embedding = self.encoder(x)
        return embedding

    def training_step(self, batch, batch_idx):
        # training_step defines the train loop. It is independent of forward
        x, _ = batch
        x = x.view(x.size(0), -1)
        z = self.encoder(x)
        x_hat = self.decoder(z)
        loss = F.mse_loss(x_hat, x)
        self.log("train_loss", loss)
        return loss

    def configure_optimizers(self):
        optimizer = torch.optim.Adam(self.parameters(), lr=1e-3)
        return optimizer


# -------------------
# Step 2: Define data
# -------------------
dataset = tv.datasets.MNIST(".", download=True, transform=tv.transforms.ToTensor())
train, val = data.random_split(dataset, [55000, 5000])

# -------------------
# Step 3: Train
# -------------------
autoencoder = LitAutoEncoder()
trainer = L.Trainer()
trainer.fit(autoencoder, data.DataLoader(train), data.DataLoader(val))

Run the model on your terminal

pip install torchvision
python main.py

Convert from PyTorch to PyTorch Lightning

PyTorch Lightning is just organized PyTorch - Lightning disentangles PyTorch code to decouple the science from the engineering.

Examples

Explore various types of training possible with PyTorch Lightning. Pretrain and finetune ANY kind of model to perform ANY task like classification, segmentation, summarization and more:

Task	Description	Run
Hello world	Pretrain - Hello world example
Image classification	Finetune - ResNet-34 model to classify images of cars
Image segmentation	Finetune - ResNet-50 model to segment images
Object detection	Finetune - Faster R-CNN model to detect objects
Text classification	Finetune - text classifier (BERT model)
Text summarization	Finetune - text summarization (Hugging Face transformer model)
Audio generation	Finetune - audio generator (transformer model)
LLM finetuning	Finetune - LLM (Meta Llama 3.1 8B)
Image generation	Pretrain - Image generator (diffusion model)
Recommendation system	Train - recommendation system (factorization and embedding)
Time-series forecasting	Train - Time-series forecasting with LSTM

Advanced features

Lightning has over 40+ advanced features designed for professional AI research at scale.

Here are some examples:

Train on 1000s of GPUs without code changes

# 8 GPUs
# no code changes needed
trainer = Trainer(accelerator="gpu", devices=8)

# 256 GPUs
trainer = Trainer(accelerator="gpu", devices=8, num_nodes=32)

Train on other accelerators like TPUs without code changes

# no code changes needed
trainer = Trainer(accelerator="tpu", devices=8)

16-bit precision

# no code changes needed
trainer = Trainer(precision=16)

Experiment managers

from lightning import loggers

# litlogger
trainer = Trainer(logger=LitLogger())

# tensorboard
trainer = Trainer(logger=TensorBoardLogger("logs/"))

# weights and biases
trainer = Trainer(logger=loggers.WandbLogger())

# comet
trainer = Trainer(logger=loggers.CometLogger())

# mlflow
trainer = Trainer(logger=loggers.MLFlowLogger())

# ... and dozens more

Early Stopping

es = EarlyStopping(monitor="val_loss")
trainer = Trainer(callbacks=[es])

Checkpointing

checkpointing = ModelCheckpoint(monitor="val_loss")
trainer = Trainer(callbacks=[checkpointing])

Export to torchscript (JIT) (production use)

# torchscript
autoencoder = LitAutoEncoder()
torch.jit.save(autoencoder.to_torchscript(), "model.pt")

Export to ONNX (production use)

# onnx
with tempfile.NamedTemporaryFile(suffix=".onnx", delete=False) as tmpfile:
    autoencoder = LitAutoEncoder()
    input_sample = torch.randn((1, 64))
    autoencoder.to_onnx(tmpfile.name, input_sample, export_params=True)
    os.path.isfile(tmpfile.name)

Advantages over unstructured PyTorch

Models become hardware agnostic
Code is clear to read because engineering code is abstracted away
Easier to reproduce
Make fewer mistakes because lightning handles the tricky engineering
Keeps all the flexibility (LightningModules are still PyTorch modules), but removes a ton of boilerplate
Lightning has dozens of integrations with popular machine learning tools.
Tested rigorously with every new PR. We test every combination of PyTorch and Python supported versions, every OS, multi GPUs and even TPUs.
Minimal running speed overhead (about 300 ms per epoch compared with pure PyTorch).

Read the PyTorch Lightning docs

Lightning Fabric: Expert control

Run on any device at any scale with expert-level control over PyTorch training loop and scaling strategy. You can even write your own Trainer.

Fabric is designed for the most complex models like foundation model scaling, LLMs, diffusion, transformers, reinforcement learning, active learning. Of any size.

What to change Resulting Fabric Code (copy me!)

_{+ import lightning as L
import torch; import torchvision as tv

dataset = tv.datasets.CIFAR10("data", download=True,
train=True,
transform=tv.transforms.ToTensor())

+ fabric = L.Fabric()
+ fabric.launch()

model = tv.models.resnet18()
optimizer = torch.optim.SGD(model.parameters(), lr=0.001)
- device = "cuda" if torch.cuda.is_available() else "cpu"
- model.to(device)
+ model, optimizer = fabric.setup(model, optimizer)

dataloader = torch.utils.data.DataLoader(dataset, batch_size=8)
+ dataloader = fabric.setup_dataloaders(dataloader)

model.train()
num_epochs = 10
for epoch in range(num_epochs):
for batch in dataloader:
inputs, labels = batch
- inputs, labels = inputs.to(device), labels.to(device)
optimizer.zero_grad()
outputs = model(inputs)
loss = torch.nn.functional.cross_entropy(outputs, labels)
- loss.backward()
+ fabric.backward(loss)
optimizer.step()
print(loss.data)}

_{import lightning as L
import torch; import torchvision as tv

dataset = tv.datasets.CIFAR10("data", download=True,
train=True,
transform=tv.transforms.ToTensor())

fabric = L.Fabric()
fabric.launch()

model = tv.models.resnet18()
optimizer = torch.optim.SGD(model.parameters(), lr=0.001)
model, optimizer = fabric.setup(model, optimizer)

dataloader = torch.utils.data.DataLoader(dataset, batch_size=8)
dataloader = fabric.setup_dataloaders(dataloader)

model.train()
num_epochs = 10
for epoch in range(num_epochs):
for batch in dataloader:
inputs, labels = batch
optimizer.zero_grad()
outputs = model(inputs)
loss = torch.nn.functional.cross_entropy(outputs, labels)
fabric.backward(loss)
optimizer.step()
print(loss.data)}

Key features

Easily switch from running on CPU to GPU (Apple Silicon, CUDA, …), TPU, multi-GPU or even multi-node training

# Use your available hardware
# no code changes needed
fabric = Fabric()

# Run on GPUs (CUDA or MPS)
fabric = Fabric(accelerator="gpu")

# 8 GPUs
fabric = Fabric(accelerator="gpu", devices=8)

# 256 GPUs, multi-node
fabric = Fabric(accelerator="gpu", devices=8, num_nodes=32)

# Run on TPUs
fabric = Fabric(accelerator="tpu")

Use state-of-the-art distributed training strategies (DDP, FSDP, DeepSpeed) and mixed precision out of the box

# Use state-of-the-art distributed training techniques
fabric = Fabric(strategy="ddp")
fabric = Fabric(strategy="deepspeed")
fabric = Fabric(strategy="fsdp")

# Switch the precision
fabric = Fabric(precision="16-mixed")
fabric = Fabric(precision="64")

All the device logic boilerplate is handled for you

  # no more of this!
- model.to(device)
- batch.to(device)

Build your own custom Trainer using Fabric primitives for training checkpointing, logging, and more

import lightning as L


class MyCustomTrainer:
    def __init__(self, accelerator="auto", strategy="auto", devices="auto", precision="32-true"):
        self.fabric = L.Fabric(accelerator=accelerator, strategy=strategy, devices=devices, precision=precision)

    def fit(self, model, optimizer, dataloader, max_epochs):
        self.fabric.launch()

        model, optimizer = self.fabric.setup(model, optimizer)
        dataloader = self.fabric.setup_dataloaders(dataloader)
        model.train()

        for epoch in range(max_epochs):
            for batch in dataloader:
                input, target = batch
                optimizer.zero_grad()
                output = model(input)
                loss = loss_fn(output, target)
                self.fabric.backward(loss)
                optimizer.step()

You can find a more extensive example in our examples

Read the Lightning Fabric docs

Examples

Self-supervised Learning

Convolutional Architectures

GPT-2
UNet

Reinforcement Learning

GANs

Classic ML

Continuous Integration

Lightning is rigorously tested across multiple CPUs, GPUs and TPUs and against major Python and PyTorch versions.

*Codecov is > 90%+ but build delays may show less

Current build statuses

System / PyTorch ver.	1.13	2.0	2.1
Linux py3.9 [GPUs]
Linux (multiple Python versions)
OSX (multiple Python versions)
Windows (multiple Python versions)

Community

The lightning community is maintained by

10+ core contributors who are all a mix of professional engineers, Research Scientists, and Ph.D. students from top AI labs.
800+ community contributors.

Want to help us build Lightning and reduce boilerplate for thousands of researchers? Learn how to make your first contribution here

Lightning is also part of the PyTorch ecosystem which requires projects to have solid testing, documentation and support.

Asking for help

If you have any questions please:

分类

Development Status :: 4 - Beta Environment :: Console Intended Audience :: Developers License :: OSI Approved :: Apache Software License Natural Language :: English Operating System :: OS Independent Programming Language :: Python :: 3 Programming Language :: Python :: 3.10 Programming Language :: Python :: 3.11 Programming Language :: Python :: 3.12 Programming Language :: Python :: 3.13 Topic :: Scientific/Engineering :: Artificial Intelligence Topic :: Scientific/Engineering :: Information Analysis

版本列表

2.6.5 2026-05-27

2.6.4 2026-05-20

2.6.1 2026-01-30

2.6.1.dev20260201 2026-02-01

2.6.0 2025-11-28

2.6.0.dev20260125 2026-01-25

2.6.0.dev20260118 2026-01-18

2.6.0.dev20260111 2026-01-11

2.6.0.dev20260104 2026-01-04

2.6.0.dev20251228 2025-12-28

2.6.0.dev20251221 2025-12-21

2.6.0.dev20251214 2025-12-14

2.6.0.dev20251207 2025-12-07

2.6.0.dev20251130 2025-11-30

2.6.0.dev20251123 2025-11-23

2.6.0.dev20251116 2025-11-16

2.6.0.dev20251109 2025-11-09

2.6.0.dev20251102 2025-11-02

2.6.0.dev20251026 2025-10-26

2.6.0.dev20251019 2025-10-19

2.6.0.dev20251012 2025-10-12

2.6.0.dev20251005 2025-10-05

2.6.0.dev20250928 2025-09-28

2.6.0.dev20250921 2025-09-21

2.6.0.dev20250914 2025-09-14

2.6.0.dev20250831 2025-08-31

2.6.0.dev20250824 2025-08-24

2.6.0.dev20250817 2025-08-17

2.6.0.dev20250810 2025-08-10

2.6.0.dev20250803 2025-08-03

2.6.0.dev20250727 2025-07-27

2.6.0.dev20250720 2025-07-20

2.6.0.dev20250713 2025-07-13

2.6.0.dev20250706 2025-07-06

2.6.0.dev20250629 2025-06-29

2.6.0.dev0 2025-11-05

2.5.6 2025-11-05

2.5.5 2025-09-05

2.5.4 2025-08-29

2.5.3 2025-08-13

2.5.2 2025-06-20

2.5.1 2025-03-19

2.5.1.post0 2025-04-25

2.5.1.dev20250622 2025-06-22

2.5.1.dev20250615 2025-06-15

2.5.1.dev20250608 2025-06-08

2.5.1.dev20250601 2025-06-01

2.5.1.dev20250525 2025-05-25

2.5.1.dev20250518 2025-05-18

2.5.1.dev20250511 2025-05-11

2.5.1.dev20250504 2025-05-04

2.5.1.dev20250427 2025-04-27

2.5.1.dev20250420 2025-04-20

2.5.1.dev20250413 2025-04-13

2.5.1.dev20250406 2025-04-06

2.5.1.dev20250330 2025-03-30

2.5.1.dev20250323 2025-03-23

2.5.0 2024-12-20

2.5.0.post0 2024-12-21

2.5.0.dev20250316 2025-03-16

2.5.0.dev20250309 2025-03-09

2.5.0.dev20250302 2025-03-02

2.5.0.dev20250223 2025-02-23

2.5.0.dev20250216 2025-02-16

2.5.0.dev20250209 2025-02-09

2.5.0.dev20250202 2025-02-02

2.5.0.dev20250126 2025-01-26

2.5.0.dev20250119 2025-01-19

2.5.0.dev20250112 2025-01-12

2.5.0.dev20250105 2025-01-05

2.5.0.dev20241229 2024-12-29

2.5.0.dev20241222 2024-12-22

2.5.0.dev20241215 2024-12-15

2.5.0.dev20241208 2024-12-08

2.5.0.dev20241201 2024-12-01

2.5.0.dev20241124 2024-11-24

2.5.0.dev20241117 2024-11-17

2.5.0.dev20241110 2024-11-10

2.5.0.dev20241103 2024-11-03

2.5.0.dev20241027 2024-10-27

2.5.1rc2 2025-03-17

2.5.1rc1 2025-03-08

2.5.1rc0 2025-03-05

2.5.0rc0 2024-12-12

2.4.0 2024-08-07

2.4.0.dev20241020 2024-10-20

2.4.0.dev20241013 2024-10-13

2.4.0.dev20241006 2024-10-06

2.4.0.dev20240929 2024-09-29

2.4.0.dev20240922 2024-09-22

2.4.0.dev20240915 2024-09-15

2.4.0.dev20240908 2024-09-08

2.4.0.dev20240901 2024-09-01

2.4.0.dev20240825 2024-08-25

2.4.0.dev20240818 2024-08-18

2.4.0.dev20240811 2024-08-11

2.4.0.dev20240804 2024-08-04

2.4.0.dev20240728 2024-07-28

2.4.0.dev20240721 2024-07-21

2.4.0.dev20240714 2024-07-14

2.4.0.dev20240707 2024-07-07

2.4.0.dev20240630 2024-06-30

2.3.3 2024-07-08

2.3.2 2024-07-04

2.3.1 2024-06-27

2.3.0 2024-06-13

2.3.0.dev20240623 2024-06-23

2.3.0.dev20240616 2024-06-16

2.3.0.dev20240609 2024-06-09

2.3.0.dev20240602 2024-06-02

2.3.0.dev20240526 2024-05-26

2.3.0.dev20240519 2024-05-19

2.3.0.dev20240505 2024-05-05

2.3.0.dev20240428 2024-04-28

2.3.0.dev20240421 2024-04-21

2.3.0.dev20240414 2024-04-14

2.3.0.dev20240407 2024-04-07

2.3.0.dev20240331 2024-03-31

2.3.0.dev20240328 2024-03-28

2.3.0.dev20240324 2024-03-24

2.3.0.dev20240318 2024-03-18

2.2.5 2024-05-22

2.2.4 2024-05-01

2.2.3 2024-04-23

2.2.2 2024-04-12

2.2.1 2024-03-04

2.2.0 2024-02-08

2.2.0.post0 2024-02-12

2.2.0rc0 2024-02-01

2.1.4 2024-02-01

2.1.3 2023-12-21

2.1.2 2023-11-15

2.1.1 2023-11-08

2.1.0 2023-10-12

2.1.0rc1 2023-10-10

2.1.0rc0 2023-08-16

2.0.9 2023-09-14

2.0.9.post0 2023-09-28

2.0.8 2023-08-30

2.0.7 2023-08-16

2.0.6 2023-07-25

2.0.5 2023-07-10

2.0.4 2023-06-22

2.0.3 2023-06-07

2.0.2 2023-04-24

2.0.1 2023-03-30

2.0.1.post0 2023-04-11

2.0.0 2023-03-15

2.0.0rc0 2023-02-23

1.9.5 2023-04-12

1.9.4 2023-03-02

1.9.3 2023-02-21

1.9.2 2023-02-15

1.9.1 2023-02-10

1.9.0 2023-01-18

1.9.0rc0 2023-01-06

1.8.6 2022-12-21

1.8.5 2022-12-15

1.8.5.post0 2022-12-16

1.8.4 2022-12-09

1.8.4.post0 2022-12-10

1.8.3 2022-11-23

1.8.3.post2 2022-12-09

1.8.3.post1 2022-11-25

1.8.3.post0 2022-11-23

1.8.2 2022-11-18

1.8.1 2022-11-10

1.8.0 2022-11-01

1.8.0.post1 2022-11-02

1.8.0rc2 2022-11-01

1.8.0rc1 2022-10-27