ddp_check_lightning.py

# /// script
# requires-python = "==3.12.12"
# dependencies = [
#   "torch==2.2.1",
#   "numpy==1.26.4",
#   "lightning==2.6.0",
#   "torchvision==0.17.1",
# ]
# ///

import os
import torch
from torch import optim, nn, utils, Tensor
from torchvision.datasets import MNIST
from torchvision.transforms import ToTensor
import lightning as L


class LitAutoEncoder(L.LightningModule):
    def __init__(self, encoder, decoder):
        super().__init__()
        self.encoder = encoder
        self.decoder = decoder

    def forward(self, x):
        # encoder only
        return self.encoder(x)

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

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

def main():
    torch.set_float32_matmul_precision("high")

    # create model
    encoder = nn.Sequential(nn.Linear(28 * 28, 64), nn.ReLU(), nn.Linear(64, 3))
    decoder = nn.Sequential(nn.Linear(3, 64), nn.ReLU(), nn.Linear(64, 28 * 28))
    model = LitAutoEncoder(encoder, decoder)
    
    # setup data
    dataset = MNIST(root=os.getcwd(), download=True, transform=ToTensor())
    train_loader = utils.data.DataLoader(dataset, batch_size=64, shuffle=True)

    # create a model and trainer
    trainer = L.Trainer(max_epochs=10, devices="auto")
    trainer.fit(model, train_loader)

if __name__ == "__main__":
    main()