您好,登錄后才能下訂單哦!
這篇“pytorch怎么實(shí)現(xiàn)加載保存查看checkpoint文件”文章的知識(shí)點(diǎn)大部分人都不太理解,所以小編給大家總結(jié)了以下內(nèi)容,內(nèi)容詳細(xì),步驟清晰,具有一定的借鑒價(jià)值,希望大家閱讀完這篇文章能有所收獲,下面我們一起來看看這篇“pytorch怎么實(shí)現(xiàn)加載保存查看checkpoint文件”文章吧。
# 方式一:保存加載整個(gè)state_dict(推薦) # 保存 torch.save(model.state_dict(), PATH) # 加載 model.load_state_dict(torch.load(PATH)) # 測(cè)試時(shí)不啟用 BatchNormalization 和 Dropout model.eval()
# 方式二:保存加載整個(gè)模型 # 保存 torch.save(model, PATH) # 加載 model = torch.load(PATH) model.eval()
# 方式三:保存用于繼續(xù)訓(xùn)練的checkpoint或者多個(gè)模型 # 保存 torch.save({ 'epoch': epoch, 'model_state_dict': model.state_dict(), ... }, PATH) # 加載 checkpoint = torch.load(PATH) start_epoch=checkpoint['epoch'] model.load_state_dict(checkpoint['model_state_dict']) # 測(cè)試時(shí) model.eval() # 或者訓(xùn)練時(shí) model.train()
# GPU上保存,CPU上加載 # 保存 torch.save(model.state_dict(), PATH) # 加載 device = torch.device('cpu') model.load_state_dict(torch.load(PATH, map_location=device)) # 如果是多gpu保存,需要去除關(guān)鍵字中的module,見第4部分
# GPU上保存,GPU上加載 # 保存 torch.save(model.state_dict(), PATH) # 加載 device = torch.device("cuda") model.load_state_dict(torch.load(PATH)) model.to(device)
# CPU上保存,GPU上加載 # 保存 torch.save(model.state_dict(), PATH) # 加載 device = torch.device("cuda") # 選擇希望使用的GPU model.load_state_dict(torch.load(PATH, map_location="cuda:0")) model.to(device)
# 打印模型的 state_dict print("Model's state_dict:") for param_tensor in model.state_dict(): print(param_tensor, "\t", model.state_dict()[param_tensor].size())
多gpu
報(bào)錯(cuò)為KeyError: ‘unexpected key “module.conv1.weight” in state_dict’
原因:當(dāng)使用多gpu時(shí),會(huì)使用torch.nn.DataParallel,所以checkpoint中有module字樣
#解決1:加載時(shí)將module去掉 # 創(chuàng)建一個(gè)不包含`module.`的新OrderedDict from collections import OrderedDict new_state_dict = OrderedDict() for k, v in state_dict.items(): name = k[7:] # 去掉 `module.` new_state_dict[name] = v # 加載參數(shù) model.load_state_dict(new_state_dict)
# 解決2:保存checkpoint時(shí)不保存module torch.save(model.module.state_dict(), PATH)
'''本文件用于舉例說明pytorch保存和加載文件的方法''' import torch as torch import torchvision as tv import torch.nn as nn import torch.optim as optim import torch.nn.functional as F import torchvision.transforms as transforms import os # 參數(shù)聲明 batch_size = 32 epochs = 10 WORKERS = 0 # dataloder線程數(shù) test_flag = False # 測(cè)試標(biāo)志,True時(shí)加載保存好的模型進(jìn)行測(cè)試 ROOT = '/home/pxt/pytorch/cifar' # MNIST數(shù)據(jù)集保存路徑 log_dir = '/home/pxt/pytorch/logs/cifar_model.pth' # 模型保存路徑 # 加載MNIST數(shù)據(jù)集 transform = tv.transforms.Compose([ transforms.ToTensor(), transforms.Normalize([0.5, 0.5, 0.5], [0.5, 0.5, 0.5])]) train_data = tv.datasets.CIFAR10(root=ROOT, train=True, download=True, transform=transform) test_data = tv.datasets.CIFAR10(root=ROOT, train=False, download=False, transform=transform) train_load = torch.utils.data.DataLoader(train_data, batch_size=batch_size, shuffle=True, num_workers=WORKERS) test_load = torch.utils.data.DataLoader(test_data, batch_size=batch_size, shuffle=False, num_workers=WORKERS) # 構(gòu)造模型 class Net(nn.Module): def __init__(self): super(Net, self).__init__() self.conv1 = nn.Conv2d(3, 64, 3, padding=1) self.conv2 = nn.Conv2d(64, 128, 3, padding=1) self.conv3 = nn.Conv2d(128, 256, 3, padding=1) self.conv4 = nn.Conv2d(256, 256, 3, padding=1) self.pool = nn.MaxPool2d(2, 2) self.fc1 = nn.Linear(256 * 8 * 8, 1024) self.fc2 = nn.Linear(1024, 256) self.fc3 = nn.Linear(256, 10) def forward(self, x): x = F.relu(self.conv1(x)) x = self.pool(F.relu(self.conv2(x))) x = F.relu(self.conv3(x)) x = self.pool(F.relu(self.conv4(x))) x = x.view(-1, x.size()[1] * x.size()[2] * x.size()[3]) x = F.relu(self.fc1(x)) x = F.relu(self.fc2(x)) x = self.fc3(x) return x model = Net().cpu() criterion = nn.CrossEntropyLoss() optimizer = optim.SGD(model.parameters(), lr=0.01) # 模型訓(xùn)練 def train(model, train_loader, epoch): model.train() train_loss = 0 for i, data in enumerate(train_loader, 0): x, y = data x = x.cpu() y = y.cpu() optimizer.zero_grad() y_hat = model(x) loss = criterion(y_hat, y) loss.backward() optimizer.step() train_loss += loss print('正在進(jìn)行第{}個(gè)epoch中的第{}次循環(huán)'.format(epoch,i)) loss_mean = train_loss / (i + 1) print('Train Epoch: {}\t Loss: {:.6f}'.format(epoch, loss_mean.item())) # 模型測(cè)試 def test(model, test_loader): model.eval() test_loss = 0 correct = 0 with torch.no_grad(): for i, data in enumerate(test_loader, 0): x, y = data x = x.cpu() y = y.cpu() optimizer.zero_grad() y_hat = model(x) test_loss += criterion(y_hat, y).item() pred = y_hat.max(1, keepdim=True)[1] correct += pred.eq(y.view_as(pred)).sum().item() test_loss /= (i + 1) print('Test set: Average loss: {:.4f}, Accuracy: {}/{} ({:.0f}%)\n'.format( test_loss, correct, len(test_data), 100. * correct / len(test_data))) def main(): # 如果test_flag=True,則加載已保存的模型并進(jìn)行測(cè)試,測(cè)試以后不進(jìn)行此模塊以后的步驟 if test_flag: # 加載保存的模型直接進(jìn)行測(cè)試機(jī)驗(yàn)證 checkpoint = torch.load(log_dir) model.load_state_dict(checkpoint['model']) optimizer.load_state_dict(checkpoint['optimizer']) start_epoch = checkpoint['epoch'] test(model, test_load) return # 如果有保存的模型,則加載模型,并在其基礎(chǔ)上繼續(xù)訓(xùn)練 if os.path.exists(log_dir): checkpoint = torch.load(log_dir) model.load_state_dict(checkpoint['model']) optimizer.load_state_dict(checkpoint['optimizer']) start_epoch = checkpoint['epoch'] print('加載 epoch {} 成功!'.format(start_epoch)) else: start_epoch = 0 print('無保存了的模型,將從頭開始訓(xùn)練!') for epoch in range(start_epoch+1, epochs): train(model, train_load, epoch) test(model, test_load) # 保存模型 state = {'model':model.state_dict(), 'optimizer':optimizer.state_dict(), 'epoch':epoch} torch.save(state, log_dir) if __name__ == '__main__': main()
以上就是關(guān)于“pytorch怎么實(shí)現(xiàn)加載保存查看checkpoint文件”這篇文章的內(nèi)容,相信大家都有了一定的了解,希望小編分享的內(nèi)容對(duì)大家有幫助,若想了解更多相關(guān)的知識(shí)內(nèi)容,請(qǐng)關(guān)注億速云行業(yè)資訊頻道。
免責(zé)聲明:本站發(fā)布的內(nèi)容(圖片、視頻和文字)以原創(chuàng)、轉(zhuǎn)載和分享為主,文章觀點(diǎn)不代表本網(wǎng)站立場(chǎng),如果涉及侵權(quán)請(qǐng)聯(lián)系站長(zhǎng)郵箱:is@yisu.com進(jìn)行舉報(bào),并提供相關(guān)證據(jù),一經(jīng)查實(shí),將立刻刪除涉嫌侵權(quán)內(nèi)容。