Terjemahan disediakan oleh mesin penerjemah. Jika konten terjemahan yang diberikan bertentangan dengan versi bahasa Inggris aslinya, utamakan versi bahasa Inggris.

Memodifikasi Skrip PyTorch Pelatihan

Di bagian ini, Anda mempelajari cara memodifikasi skrip PyTorch pelatihan untuk mengonfigurasi pustaka paralelisme SageMaker model untuk partisi otomatis dan partisi manual.

Perhatikan bahwa partisi otomatis diaktifkan secara default. Kecuali ditentukan lain, skrip berikut menggunakan partisi otomatis.

Pemisahan otomatis dengan PyTorch

Perubahan skrip pelatihan berikut diperlukan untuk menjalankan skrip PyTorch pelatihan dengan pustaka SageMaker paralelisme model:

Untuk mempelajari lebih lanjut tentang API pustaka paralelisme model, lihat dokumentasi API. SageMaker

import torch
import torch.nn as nn
import torch.nn.functional as F
import torch.optim as optim
from torchnet.dataset import SplitDataset
from torchvision import datasets

import smdistributed.modelparallel.torch as smp

class GroupedNet(nn.Module):
    def __init__(self):
        super(GroupedNet, self).__init__()
        # define layers

    def forward(self, x):
        # define forward pass and return model outputs


# smdistributed: Define smp.step. Return any tensors needed outside.
@smp.step
def train_step(model, data, target):
    output = model(data)
    loss = F.nll_loss(output, target, reduction="mean")
    model.backward(loss)
    return output, loss


def train(model, device, train_loader, optimizer):
    model.train()
    for batch_idx, (data, target) in enumerate(train_loader):
        # smdistributed: Move input tensors to the GPU ID used by the current process,
        # based on the set_device call.
        data, target = data.to(device), target.to(device)
        optimizer.zero_grad()
        # Return value, loss_mb is a StepOutput object
        _, loss_mb = train_step(model, data, target)

        # smdistributed: Average the loss across microbatches.
        loss = loss_mb.reduce_mean()

        optimizer.step()

# smdistributed: initialize the backend
smp.init()

# smdistributed: Set the device to the GPU ID used by the current process.
# Input tensors should be transferred to this device.
torch.cuda.set_device(smp.local_rank())
device = torch.device("cuda")

# smdistributed: Download only on a single process per instance.
# When this is not present, the file is corrupted by multiple processes trying
# to download and extract at the same time
dataset = datasets.MNIST("../data", train=True, download=False)

# smdistributed: Shard the dataset based on data-parallel ranks
if smp.dp_size() > 1:
    partitions_dict = {f"{i}": 1 / smp.dp_size() for i in range(smp.dp_size())}
    dataset = SplitDataset(dataset, partitions=partitions_dict)
    dataset.select(f"{smp.dp_rank()}")

# smdistributed: Set drop_last=True to ensure that batch size is always divisible
# by the number of microbatches
train_loader = torch.utils.data.DataLoader(dataset, batch_size=64, drop_last=True)

model = GroupedNet()
optimizer = optim.Adadelta(model.parameters(), lr=4.0)

# smdistributed: Use the DistributedModel container to provide the model
# to be partitioned across different ranks. For the rest of the script,
# the returned DistributedModel object should be used in place of
# the model provided for DistributedModel class instantiation.
model = smp.DistributedModel(model)
optimizer = smp.DistributedOptimizer(optimizer)

train(model, device, train_loader, optimizer)

Pemisahan manual dengan PyTorch

Gunakan manajer smp.partitionkonteks untuk menempatkan modul di perangkat tertentu. Modul apa pun yang tidak ditempatkan dalam smp.partition konteks apa pun ditempatkan di. default_partition default_partitionKebutuhan yang harus auto_partition disediakan jika diatur keFalse. Modul yang dibuat dalam smp.partition konteks tertentu ditempatkan pada partisi yang sesuai.

Untuk mempelajari lebih lanjut tentang API pustaka paralelisme model, lihat dokumentasi API. SageMaker

import torch
import torch.nn as nn
import torch.nn.functional as F
import torch.optim as optim
from torchnet.dataset import SplitDataset
from torchvision import datasets

import smdistributed.modelparallel.torch as smp

class GroupedNet(nn.Module):
    def __init__(self):
        super(GroupedNet, self).__init__()
        with smp.partition(0):
            # define child modules on device 0
        with smp.partition(1):
            # define child modules on device 1

    def forward(self, x):
        # define forward pass and return model outputs


# smdistributed: Define smp.step. Return any tensors needed outside.
@smp.step
def train_step(model, data, target):
    output = model(data)
    loss = F.nll_loss(output, target, reduction="mean")
    model.backward(loss)
    return output, loss


def train(model, device, train_loader, optimizer):
    model.train()
    for batch_idx, (data, target) in enumerate(train_loader):
        # smdistributed: Move input tensors to the GPU ID used by the current process,
        # based on the set_device call.
        data, target = data.to(device), target.to(device)
        optimizer.zero_grad()
        # Return value, loss_mb is a StepOutput object
        _, loss_mb = train_step(model, data, target)

        # smdistributed: Average the loss across microbatches.
        loss = loss_mb.reduce_mean()

        optimizer.step()

# smdistributed: initialize the backend
smp.init()

# smdistributed: Set the device to the GPU ID used by the current process.
# Input tensors should be transferred to this device.
torch.cuda.set_device(smp.local_rank())
device = torch.device("cuda")

# smdistributed: Download only on a single process per instance.
# When this is not present, the file is corrupted by multiple processes trying
# to download and extract at the same time
dataset = datasets.MNIST("../data", train=True, download=False)

# smdistributed: Shard the dataset based on data-parallel ranks
if smp.dp_size() > 1:
    partitions_dict = {f"{i}": 1 / smp.dp_size() for i in range(smp.dp_size())}
    dataset = SplitDataset(dataset, partitions=partitions_dict)
    dataset.select(f"{smp.dp_rank()}")

# smdistributed: Set drop_last=True to ensure that batch size is always divisible
# by the number of microbatches
train_loader = torch.utils.data.DataLoader(dataset, batch_size=64, drop_last=True)

model = GroupedNet()
optimizer = optim.Adadelta(model.parameters(), lr=4.0)

# smdistributed: Use the DistributedModel container to provide the model
# to be partitioned across different ranks. For the rest of the script,
# the returned DistributedModel object should be used in place of
# the model provided for DistributedModel class instantiation.
model = smp.DistributedModel(model)
optimizer = smp.DistributedOptimizer(optimizer)

train(model, device, train_loader, optimizer)

Pertimbangan

Saat Anda mengonfigurasi skrip PyTorch pelatihan menggunakan SageMaker pustaka paralelisme model, Anda harus mengetahui hal berikut:

Fitur kerangka kerja yang tidak didukung

PyTorch Fitur-fitur berikut tidak didukung oleh pustaka SageMaker paralelisme model: