an example: pytorch to caffe2

mnist_pytorch2caffe2.py

from __future__ import print_function
import argparse
from datetime import datetime as dt

import numpy as np

import onnx_caffe2.backend

import torch
import torch.nn as nn
import torch.nn.functional as F
import torch.onnx
import torch.optim as optim
from torchvision import datasets
from torchvision import transforms
from torch.autograd import Variable

import onnx


# Training settings
parser = argparse.ArgumentParser(description='PyTorch MNIST Example')
parser.add_argument('--batch-size', type=int, default=64, metavar='N',
                    help='input batch size for training (default: 64)')
parser.add_argument('--test-batch-size', type=int, default=1000, metavar='N',
                    help='input batch size for testing (default: 1000)')
parser.add_argument('--epochs', type=int, default=10, metavar='N',
                    help='number of epochs to train (default: 10)')
parser.add_argument('--lr', type=float, default=0.01, metavar='LR',
                    help='learning rate (default: 0.01)')
parser.add_argument('--momentum', type=float, default=0.5, metavar='M',
                    help='SGD momentum (default: 0.5)')
parser.add_argument('--no-cuda', action='store_true', default=False,
                    help='disables CUDA training')
parser.add_argument('--seed', type=int, default=1, metavar='S',
                    help='random seed (default: 1)')
parser.add_argument('--log-interval', type=int, default=10, metavar='N',
                    help='how many batches to wait before logging training status')
args = parser.parse_args()
args.cuda = not args.no_cuda and torch.cuda.is_available()

torch.manual_seed(args.seed)
if args.cuda:
    torch.cuda.manual_seed(args.seed)


kwargs = {'num_workers': 1, 'pin_memory': True} if args.cuda else {}
train_loader = torch.utils.data.DataLoader(
    datasets.MNIST('../data', train=True, download=True,
                   transform=transforms.Compose([
                       transforms.ToTensor(),
                       transforms.Normalize((0.1307,), (0.3081,))
                   ])),
    batch_size=args.batch_size, shuffle=True, **kwargs)
test_loader = torch.utils.data.DataLoader(
    datasets.MNIST('../data', train=False, transform=transforms.Compose([
                       transforms.ToTensor(),
                       transforms.Normalize((0.1307,), (0.3081,))
                   ])),
    batch_size=args.test_batch_size, shuffle=True, **kwargs)


class Net(nn.Module):
    def __init__(self):
        super(Net, self).__init__()
        self.conv1 = nn.Conv2d(1, 10, kernel_size=5)
        self.conv2 = nn.Conv2d(10, 20, kernel_size=5)
        self.conv2_drop = nn.Dropout2d()
        self.fc1 = nn.Linear(320, 50)
        self.fc2 = nn.Linear(50, 10)

    def forward(self, x):
        x = F.relu(F.max_pool2d(self.conv1(x), 2))
        x = F.relu(F.max_pool2d(self.conv2_drop(self.conv2(x)), 2))
        x = x.view(-1, 320)
        x = F.relu(self.fc1(x))
        x = F.dropout(x, training=self.training)
        x = self.fc2(x)
        return F.log_softmax(x)

model = Net()
if args.cuda:
    model.cuda()

optimizer = optim.SGD(model.parameters(), lr=args.lr, momentum=args.momentum)

def train(epoch):
    model.train()
    for batch_idx, (data, target) in enumerate(train_loader):
        if args.cuda:
            data, target = data.cuda(), target.cuda()
        data, target = Variable(data), Variable(target)
        optimizer.zero_grad()
        output = model(data)
        loss = F.nll_loss(output, target)
        loss.backward()
        optimizer.step()
        if batch_idx % args.log_interval == 0:
            print('Train Epoch: {} [{}/{} ({:.0f}%)]\tLoss: {:.6f}'.format(
                epoch, batch_idx * len(data), len(train_loader.dataset),
                100. * batch_idx / len(train_loader), loss.data[0]))

def test():
    model.eval()
    test_loss = 0
    correct = 0
    for data, target in test_loader:
        if args.cuda:
            data, target = data.cuda(), target.cuda()
        data, target = Variable(data, volatile=True), Variable(target)
        output = model(data)
        test_loss += F.nll_loss(output, target, size_average=False).data[0] # sum up batch loss
        pred = output.data.max(1, keepdim=True)[1] # get the index of the max log-probability
        correct += pred.eq(target.data.view_as(pred)).cpu().sum()

    test_loss /= len(test_loader.dataset)
    print('\nTest set: Average loss: {:.4f}, Accuracy: {}/{} ({:.0f}%)\n'.format(
        test_loss, correct, len(test_loader.dataset),
        100. * correct / len(test_loader.dataset)))


for epoch in range(1, args.epochs + 1):
    train(epoch)
    test()

print('# Export trained model!')
dummy_input = Variable(torch.randn(64, 1, 28, 28))
if args.cuda:
    dummy_input = dummy_input.cuda()
model.train(False)
torch.onnx.export(model, dummy_input, 'trained.proto', verbose=True)
print('# Load exported model via onnx')
onnx_model = onnx.load("trained.proto")
onnx.helper.printable_graph(onnx_model.graph)
print('# copy to caffe2')
device = 'CUDA:0' if args.cuda else 'CPU'
rep = onnx_caffe2.backend.prepare(onnx_model, device=device)

inputs = np.random.normal(size=(64, 1, 28, 28)).astype(np.float32)
pytorch_timer = list()
for _ in range(10):
    pytorch_start = dt.now()
    outputs = model(dummy_input)
    pytorch_end = dt.now()
    pytorch_timer.append((pytorch_end - pytorch_start).total_seconds())
caffe2_timer = list()
for _ in range(10):
    caffe2_start = dt.now()
    outputs = rep.run(inputs)
    caffe2_end = dt.now()
    caffe2_timer.append((caffe2_end - caffe2_start).total_seconds())
print('PyTorch')
print(np.mean(pytorch_timer))
print(np.std(pytorch_timer))
print('Caffe2')
print(np.mean(caffe2_timer))
print(np.std(caffe2_timer))