使用TPU在PyTorch中實現ResNet50

做者|DR. VAIBHAV KUMAR
編譯|VK
來源|Analytics In Diamag

PyTorch經過提供大量強大的工具和技術，一直在推進計算機視覺和深度學習領域的發展。

在計算機視覺領域，基於深度學習的執行須要處理大量的圖像數據集，所以須要一個加速的環境來加快執行過程以達到可接受的精度水平。

PyTorch經過XLA（加速線性代數）提供了這一特性，XLA是一種線性代數編譯器，能夠針對多種類型的硬件，包括GPU和TPU。PyTorch/XLA環境與Google雲TPU集成，實現了更快的執行速度。

在本文中，咱們將在PyTorch中使用TPU演示一種深卷積神經網絡ResNet50的實現。

該模型將在PyTorch/XLA環境中進行訓練和測試，以完成CIFAR10數據集的分類任務。咱們還將檢查在50個epoch訓練所花費的時間。

ResNet50在Pytorch的實現

爲了利用TPU的功能，這個實現是在Google Colab中完成的。首先，咱們須要從Notebook設置下的硬件加速器中選擇TPU。

選擇TPU後，咱們將使用下面的行驗證環境代碼：

import os
assert os.environ['COLAB_TPU_ADDR']

若是啓用了TPU，它將成功執行，不然它將拋出‘KeyError: ‘COLAB_TPU_ADDR’’。你也能夠經過打印TPU地址來檢查TPU。

TPU_Path = 'grpc://'+os.environ['COLAB_TPU_ADDR']
print('TPU Address:', TPU_Path)

在下一步中，咱們將安裝XLA環境以加快執行過程。咱們在上一篇文章中實現了卷積神經網絡。

VERSION = "20200516"
!curl https://raw.githubusercontent.com/pytorch/xla/master/contrib/scripts/env-setup.py -o pytorch-xla-env-setup.py
!python pytorch-xla-env-setup.py --version $VERSION

如今，咱們將在這裏導入全部必需的庫。

from matplotlib import pyplot as plt
import numpy as np
import os
import time
import torch
import torch.nn as nn
import torch.nn.functional as F
import torch.optim as optim
import torch_xla
import torch_xla.core.xla_model as xm
import torch_xla.debug.metrics as met
import torch_xla.distributed.parallel_loader as pl
import torch_xla.distributed.xla_multiprocessing as xmp
import torch_xla.utils.utils as xu
import torchvision
from torchvision import datasets, transforms
import time
from google.colab.patches import cv2_imshow
import cv2

導入庫以後，咱們將定義並初始化所需的參數。

# 定義參數
FLAGS = {}
FLAGS['data_dir'] = "/tmp/cifar"
FLAGS['batch_size'] = 128
FLAGS['num_workers'] = 4
FLAGS['learning_rate'] = 0.02
FLAGS['momentum'] = 0.9
FLAGS['num_epochs'] = 50
FLAGS['num_cores'] = 8
FLAGS['log_steps'] = 20
FLAGS['metrics_debug'] = False

在下一步中，咱們將定義ResNet50模型。

class BasicBlock(nn.Module):
  expansion = 1

  def __init__(self, in_planes, planes, stride=1):
    super(BasicBlock, self).__init__()
    self.conv1 = nn.Conv2d(
        in_planes, planes, kernel_size=3, stride=stride, padding=1, bias=False)
    self.bn1 = nn.BatchNorm2d(planes)
    self.conv2 = nn.Conv2d(
        planes, planes, kernel_size=3, stride=1, padding=1, bias=False)
    self.bn2 = nn.BatchNorm2d(planes)

    self.shortcut = nn.Sequential()
    if stride != 1 or in_planes != self.expansion * planes:
      self.shortcut = nn.Sequential(
          nn.Conv2d(
              in_planes,
              self.expansion * planes,
              kernel_size=1,
              stride=stride,
              bias=False), nn.BatchNorm2d(self.expansion * planes))

  def forward(self, x):
    out = F.relu(self.bn1(self.conv1(x)))
    out = self.bn2(self.conv2(out))
    out += self.shortcut(x)
    out = F.relu(out)
    return out

class ResNet(nn.Module):

  def __init__(self, block, num_blocks, num_classes=10):
    super(ResNet, self).__init__()
    self.in_planes = 64

    self.conv1 = nn.Conv2d(
        3, 64, kernel_size=3, stride=1, padding=1, bias=False)
    self.bn1 = nn.BatchNorm2d(64)
    self.layer1 = self._make_layer(block, 64, num_blocks[0], stride=1)
    self.layer2 = self._make_layer(block, 128, num_blocks[1], stride=2)
    self.layer3 = self._make_layer(block, 256, num_blocks[2], stride=2)
    self.layer4 = self._make_layer(block, 512, num_blocks[3], stride=2)
    self.linear = nn.Linear(512 * block.expansion, num_classes)

  def _make_layer(self, block, planes, num_blocks, stride):
    strides = [stride] + [1] * (num_blocks - 1)
    layers = []
    for stride in strides:
      layers.append(block(self.in_planes, planes, stride))
      self.in_planes = planes * block.expansion
    return nn.Sequential(*layers)

  def forward(self, x):
    out = F.relu(self.bn1(self.conv1(x)))
    out = self.layer1(out)
    out = self.layer2(out)
    out = self.layer3(out)
    out = self.layer4(out)
    out = F.avg_pool2d(out, 4)
    out = torch.flatten(out, 1)
    out = self.linear(out)
    return F.log_softmax(out, dim=1)

def ResNet50():
  return ResNet(BasicBlock, [3, 4, 6, 4, 3])

下面的代碼片斷將定義加載CIFAR10數據集、準備訓練和測試數據集、訓練過程和測試過程的函數。

SERIAL_EXEC = xmp.MpSerialExecutor()
# 只在內存中實例化一次模型權重。
WRAPPED_MODEL = xmp.MpModelWrapper(ResNet50())

def train_resnet50():
  torch.manual_seed(1)

  def get_dataset():
    norm = transforms.Normalize(
        mean=(0.4914, 0.4822, 0.4465), std=(0.2023, 0.1994, 0.2010))
    transform_train = transforms.Compose([
        transforms.RandomCrop(32, padding=4),
        transforms.RandomHorizontalFlip(),
        transforms.ToTensor(),
        norm,
    ])
    transform_test = transforms.Compose([
        transforms.ToTensor(),
        norm,
    ])
    train_dataset = datasets.CIFAR10(
        root=FLAGS['data_dir'],
        train=True,
        download=True,
        transform=transform_train)
    test_dataset = datasets.CIFAR10(
        root=FLAGS['data_dir'],
        train=False,
        download=True,
        transform=transform_test)
   
    return train_dataset, test_dataset
 
  # 使用串行執行器能夠避免多個進程
  # 下載相同的數據。
  train_dataset, test_dataset = SERIAL_EXEC.run(get_dataset)

  train_sampler = torch.utils.data.distributed.DistributedSampler(
      train_dataset,
      num_replicas=xm.xrt_world_size(),
      rank=xm.get_ordinal(),
      shuffle=True)
  train_loader = torch.utils.data.DataLoader(
      train_dataset,
      batch_size=FLAGS['batch_size'],
      sampler=train_sampler,
      num_workers=FLAGS['num_workers'],
      drop_last=True)
  test_loader = torch.utils.data.DataLoader(
      test_dataset,
      batch_size=FLAGS['batch_size'],
      shuffle=False,
      num_workers=FLAGS['num_workers'],
      drop_last=True)

  # 將學習率縮放
  learning_rate = FLAGS['learning_rate'] * xm.xrt_world_size()

  # 獲取損失函數、優化器和模型
  device = xm.xla_device()
  model = WRAPPED_MODEL.to(device)
  optimizer = optim.SGD(model.parameters(), lr=learning_rate,
                        momentum=FLAGS['momentum'], weight_decay=5e-4)
  loss_fn = nn.NLLLoss()

  def train_loop_fn(loader):
    tracker = xm.RateTracker()
    model.train()
    for x, (data, target) in enumerate(loader):
      optimizer.zero_grad()
      output = model(data)
      loss = loss_fn(output, target)
      loss.backward()
      xm.optimizer_step(optimizer)
      tracker.add(FLAGS['batch_size'])
      if x % FLAGS['log_steps'] == 0:
        print('[xla:{}]({}) Loss={:.2f} Time={}'.format(xm.get_ordinal(), x, loss.item(), time.asctime()), flush=True)

  def test_loop_fn(loader):
    total_samples = 0
    correct = 0
    model.eval()
    data, pred, target = None, None, None
    for data, target in loader:
      output = model(data)
      pred = output.max(1, keepdim=True)[1]
      correct += pred.eq(target.view_as(pred)).sum().item()
      total_samples += data.size()[0]

    accuracy = 100.0 * correct / total_samples
    print('[xla:{}] Accuracy={:.2f}%'.format(
        xm.get_ordinal(), accuracy), flush=True)
    return accuracy, data, pred, target

  # 訓練和評估的循環
  accuracy = 0.0
  data, pred, target = None, None, None
  for epoch in range(1, FLAGS['num_epochs'] + 1):
    para_loader = pl.ParallelLoader(train_loader, [device])
    train_loop_fn(para_loader.per_device_loader(device))
    xm.master_print("Finished training epoch {}".format(epoch))

    para_loader = pl.ParallelLoader(test_loader, [device])
    accuracy, data, pred, target  = test_loop_fn(para_loader.per_device_loader(device))
    if FLAGS['metrics_debug']:
      xm.master_print(met.metrics_report(), flush=True)

  return accuracy, data, pred, target

如今，咱們將開始ResNet50的訓練。訓練將在咱們在參數中定義的50個epoch內完成。訓練開始前，咱們會記錄訓練時間，訓練結束後，咱們將打印總時間。

start_time = time.time()
# 啓動訓練流程
def training(rank, flags):
  global FLAGS
  FLAGS = flags
  torch.set_default_tensor_type('torch.FloatTensor')
  accuracy, data, pred, target = train_resnet50()
  if rank == 0:
    # 檢索TPU核心0上的張量並繪製。
    plot_results(data.cpu(), pred.cpu(), target.cpu())

xmp.spawn(training, args=(FLAGS,), nprocs=FLAGS['num_cores'],
          start_method='fork')

訓練結束後，咱們會打印訓練過程所花費的時間。

最後，在訓練過程當中，咱們將模型對樣本測試數據的預測可視化。

end_time = time.time()
print("Time taken = ", end_time-start_time)

原文連接：https://analyticsindiamag.com/hands-on-guide-to-implement-resnet50-in-pytorch-with-tpu/

歡迎關注磐創AI博客站：
http://panchuang.net/

sklearn機器學習中文官方文檔：
http://sklearn123.com/

歡迎關注磐創博客資源彙總站：
http://docs.panchuang.net/