Fu_Xingwen
Fu_Xingwen

WRN28_04 on CIFAR100多显卡协同训练

时间20210502
要是有帮助的话，记得点赞收藏加关注嗷！
作者：知道许多的橘子
实现：WRN28_04对CIFAR-100数据集的分类
测试集准确度：76%
实现框架pytorch
数据增强方法：Normalize+Fix等
训练次数：200
阶段学习率[0-200]：smooth_step(10,40,100,150,epoch_s)
优化器optimizer = torch.optim.SGD(model.parameters(),lr=smooth_step(10,40,100,150,epoch_s), momentum=0.9,weight_decay=1e-5)
如果感觉算力不够用了，或者心疼自己电脑了！
可以用我实验室的算力，试试呢！
害，谁叫我的算力都用不完呢！
支持所有框架！实际上框架都配置好了！
傻瓜式云计算！
Tesla v100 1卡，2卡，4卡，8卡
内存16-128G
cpu：8-24核
想要？加个微信：15615634293
欢迎打扰！
# In[1] 导入所需工具包
from __future__ import division
import os
import torch
import torch.nn as nn
import torchvision
from torchvision import datasets,transforms
import torch.utils.data as Data
import time
from torch.nn import functional as F
from math import floor, ceil
import math
import numpy as np
import sys
sys.path.append(r'/home/megstudio/workspace/')
from FMIX.fmix import sample_and_apply, sample_mask
#import torchvision.transforms as transforms
device = torch.device('cuda' if torch.cuda.is_available() else 'cpu')
print(device)
import random

from torch.autograd import Variable
# In[1] 设置超参数
num_epochs = 200
batch_size = 100
tbatch_size = 100
test_name = '/D_W28_04_BJGB3_BLN_S_C100'

class RandomErasing(object):
    '''
    Class that performs Random Erasing in Random Erasing Data Augmentation by Zhong et al.
    -------------------------------------------------------------------------------------
    probability: The probability that the operation will be performed.
    sl: min erasing area
    sh: max erasing area
    r1: min aspect ratio
    mean: erasing value
    -------------------------------------------------------------------------------------
    '''
    def __init__(self, probability = 0.5, sl = 0.02, sh = 0.4, r1 = 0.3, mean=[0.4914, 0.4822, 0.4465]):
        self.probability = probability
        self.mean = mean
        self.sl = sl
        self.sh = sh
        self.r1 = r1

    def __call__(self, img):

        if random.uniform(0, 1) > self.probability:
            return img

        for attempt in range(100):
            area = img.size()[1] * img.size()[2]

            target_area = random.uniform(self.sl, self.sh) * area
            aspect_ratio = random.uniform(self.r1, 1/self.r1)

            h = int(round(math.sqrt(target_area * aspect_ratio)))
            w = int(round(math.sqrt(target_area / aspect_ratio)))

            if w <= img.size()[2] and h <= img.size()[1]:
                x1 = random.randint(0, img.size()[1] - h)
                y1 = random.randint(0, img.size()[2] - w)
                if img.size()[0] == 3:
                    img[0, x1:x1+h, y1:y1+w] = self.mean[0]
                    img[1, x1:x1+h, y1:y1+w] = self.mean[1]
                    img[2, x1:x1+h, y1:y1+w] = self.mean[2]
                else:
                    img[0, x1:x1+h, y1:y1+w] = self.mean[0]
                return img

        return img

# In[1] 加载数据
# 定义对数据的预处理
train_transform = transforms.Compose([
            transforms.RandomCrop(32, padding=4),  # 先四周填充0，在吧图像随机裁剪成32*32
            transforms.RandomHorizontalFlip(),  # 图像一半的概率翻转，一半的概率不翻转
            transforms.ToTensor(),
            transforms.Normalize((0.4914, 0.4822, 0.4465), (0.2023, 0.1994, 0.2010)),
            RandomErasing(probability = 0.5, sl = 0.02, sh = 0.4, r1 = 0.3, mean=[0.4914, 0.4822, 0.4465])])

test_transform = transforms.Compose([
            transforms.ToTensor(),
            transforms.Normalize((0.4914, 0.4822, 0.4465), (0.2023, 0.1994, 0.2010))])

# 训练集
trainset = torchvision.datasets.CIFAR100(
                    root='/home/megstudio/dataset/dataset-2097/file-1251/CIFAR100',
                    train=True,
                    download=False,
                    transform=train_transform)
#print(type(trainset.train_data))
train_loader = Data.DataLoader(
                    dataset=trainset,
                    batch_size=batch_size,
                    shuffle=True,
                    num_workers=2)



#得到测试集的数据
testset = torchvision.datasets.CIFAR100(
                    root='/home/megstudio/dataset/dataset-2097/file-1251/CIFAR100',
                    train=False,
                    download=False,
                    transform=test_transform)

test_loader = Data.DataLoader(
                    dataset=testset,
                    batch_size=tbatch_size,
                    shuffle=False,
                    num_workers=2)

# In[1] 加载模型

# Code modified from https://github.com/xternalz/WideResNet-pytorch

class BasicBlock(nn.Module):
    def __init__(self, in_planes, out_planes, stride, dropRate=0.0):
        super(BasicBlock, self).__init__()
        self.bn1 = nn.BatchNorm2d(in_planes)
        self.relu1 = nn.ReLU(inplace=True)
        self.conv1 = nn.Conv2d(in_planes, out_planes, kernel_size=3, stride=stride,
                               padding=1, bias=False)
        self.bn2 = nn.BatchNorm2d(out_planes)
        self.relu2 = nn.ReLU(inplace=True)
        self.conv2 = nn.Conv2d(out_planes, out_planes, kernel_size=3, stride=1,
                               padding=1, bias=False)
        self.droprate = dropRate
        self.equalInOut = (in_planes == out_planes)
        self.convShortcut = (not self.equalInOut) and nn.Conv2d(in_planes, out_planes, kernel_size=1, stride=stride,
                                                                padding=0, bias=False) or None

    def forward(self, x):
        if not self.equalInOut:
            x = self.relu1(self.bn1(x))
        else:
            out = self.relu1(self.bn1(x))
        out = self.relu2(self.bn2(self.conv1(out if self.equalInOut else x)))
        if self.droprate > 0:
            out = F.dropout(out, p=self.droprate, training=self.training)
        out = self.conv2(out)
        return torch.add(x if self.equalInOut else self.convShortcut(x), out)


class NetworkBlock(nn.Module):
    def __init__(self, nb_layers, in_planes, out_planes, block, stride, dropRate=0.0):
        super(NetworkBlock, self).__init__()
        self.layer = self._make_layer(block, in_planes, out_planes, nb_layers, stride, dropRate)

    def _make_layer(self, block, in_planes, out_planes, nb_layers, stride, dropRate):
        layers = []
        for i in range(nb_layers):
            layers.append(block(i == 0 and in_planes or out_planes, out_planes, i == 0 and stride or 1, dropRate))
        return nn.Sequential(*layers)

    def forward(self, x):
        return self.layer(x)


class WideResNet(nn.Module):
    def __init__(self, depth, num_classes, widen_factor=1, dropRate=0.0, nc=1):
        super(WideResNet, self).__init__()
        nChannels = [16, 16 * widen_factor, 32 * widen_factor, 64 * widen_factor]
        assert (depth - 4) % 6 == 0, 'depth should be 6n+4'
        n = (depth - 4) // 6
        block = BasicBlock
        # 1st conv before any network block
        self.conv1 = nn.Conv2d(nc, nChannels[0], kernel_size=3, stride=1,
                               padding=1, bias=False)
        # 1st block
        self.block1 = NetworkBlock(n, nChannels[0], nChannels[1], block, 1, dropRate)
        # 2nd block
        self.block2 = NetworkBlock(n, nChannels[1], nChannels[2], block, 2, dropRate)
        # 3rd block
        self.block3 = NetworkBlock(n, nChannels[2], nChannels[3], block, 2, dropRate)
        # global average pooling and classifier
        self.bn1 = nn.BatchNorm2d(nChannels[3])
        self.relu = nn.ReLU(inplace=True)
        self.fc = nn.Linear(nChannels[3], num_classes)
        self.nChannels = nChannels[3]

        for m in self.modules():
            if isinstance(m, nn.Conv2d):
                n = m.kernel_size[0] * m.kernel_size[1] * m.out_channels
                m.weight.data.normal_(0, math.sqrt(2. / n))
            elif isinstance(m, nn.BatchNorm2d):
                m.weight.data.fill_(1)
                m.bias.data.zero_()
            elif isinstance(m, nn.Linear):
                m.bias.data.zero_()

    def forward(self, x):
        out = self.conv1(x)
        out = self.block1(out)
        out = self.block2(out)
        out = self.block3(out)
        out = self.relu(self.bn1(out))
        out = F.avg_pool2d(out, 7)
        out = out.view(-1, self.nChannels)
        return self.fc(out)


def wrn(**kwargs):
    """
    Constructs a Wide Residual Networks.
    """
    model = WideResNet(**kwargs)
    return model


def smooth_step(a,b,c,d,x):
    level_s=0.01
    level_m=0.1
    level_n=0.01
    level_r=0.005
    if x<=a:
        return level_s
    if a<x<=b:
        return (((x-a)/(b-a))*(level_m-level_s)+level_s)
    if b<x<=c:
        return level_m
    if c<x<=d:
        return level_n
    if d<x:
        return level_r
# In[1] 设置一个通过优化器更新学习率的函数
def update_lr(optimizer, lr):
    for param_group in optimizer.param_groups:
        param_group['lr'] = lr
# In[1] 定义测试函数
def test(model,test_loader):
    
    model.eval()
    with torch.no_grad():
        correct = 0
        total = 0
        for images, labels in test_loader:
            images = images.to(device)
            labels = labels.to(device)
            outputs= model(images)

            _, predicted = torch.max(outputs.data, 1)

            total += labels.size(0)
            correct += (predicted == labels).sum().item()
        acc=100 * correct / total
    
        print('Accuracy of the model on the test images: {} %'.format(acc))
    return acc

# In[1] 定义模型和损失函数
# =============================================================================
def mkdir(path):
    folder = os.path.exists(path)
    if not folder:                   #判断是否存在文件夹如果不存在则创建为文件夹
        os.makedirs(path)            #makedirs 创建文件时如果路径不存在会创建这个路径
        print("---  new folder...  ---")
        print("---  OK  ---")
    else:
        print("---  There is this folder!  ---")
path = os.getcwd()
path = path+test_name
print(path)
mkdir(path)             #调用函数
# In[1] 定义模型和损失函数
#$$
# =============================================================================
try:
    model = torch.load(path+'/model.pkl').to(device)
    epoch_s = np.load(path+'/learning_rate.npy')
    #learning_rate *= 3
    print(epoch_s)
    train_loss = np.load(path+'/test_acc.npy').tolist()
    test_acc = np.load(path+'/test_acc.npy').tolist()
    print("---  There is a model in the folder...  ---")
except:
    print("---  Create a new model...  ---")
    epoch_s = 0
    model = wrn(depth=28, num_classes=100, widen_factor=4, dropRate=0.4, nc=3).to(device)
    train_loss=[]#准备放误差     
    test_acc=[]#准备放测试准确率
# =============================================================================
#多卡协同工作
if torch.cuda.device_count() > 1:
    model = nn.DataParallel(model)
    print(torch.cuda.device_count())
# =============================================================================
def saveModel(model,epoch,test_acc,train_loss):
    torch.save(model, path+'/model.pkl')
    # torch.save(model.state_dict(), 'resnet.ckpt')
    epoch_save=np.array(epoch)
    np.save(path+'/learning_rate.npy',epoch_save)
    test_acc=np.array(test_acc)
    np.save(path+'/test_acc.npy',test_acc)
    train_loss=np.array(train_loss)
    np.save(path+'/train_loss.npy',train_loss)
    
criterion = nn.CrossEntropyLoss()
#optimizer = torch.optim.Adam(model.parameters(), lr=learning_rate, weight_decay=1e-5)
optimizer = torch.optim.SGD(model.parameters(),lr=smooth_step(10,40,100,150,epoch_s), momentum=0.9,weight_decay=1e-5) 
#scheduler = torch.optim.lr_scheduler.MultiStepLR(optimizer,  milestones = [100, 150], gamma = 0.1, last_epoch=-1)
# In[1] 训练模型更新学习率
total_step = len(train_loader)
#curr_lr = learning_rate
for epoch in range(epoch_s, num_epochs):
    #scheduler.step()
    in_epoch = time.time()
    for i, (images, labels) in enumerate(train_loader):
        
        # =============================================================================
        # FMiX
        # print(type(images))
        images, index, lam = sample_and_apply(images, alpha=1, decay_power=3, shape=(32,32))
        images = images.type(torch.FloatTensor) 
        shuffled_label = labels[index].to(device)
        
        images = images.to(device)
        labels = labels.to(device)
        # Forward pass
        outputs = model(images)
        loss = lam*criterion(outputs, labels) + (1-lam)*criterion(outputs, shuffled_label)
        # loss = criterion(outputs, labels)
        # =============================================================================

        # Backward and optimize
        optimizer.zero_grad()
        loss.backward(retain_graph=False)
        optimizer.step()
        
        if (i + 1) % 100 == 0:
            print("Epoch [{}/{}], Step [{}/{}] Loss: {:.4f}"
                  .format(epoch + 1, num_epochs, i + 1, total_step, loss.item()))
    # 记录误差和精度
    train_loss.append(loss.item())
    acctemp = test(model, test_loader)
    test_acc.append(acctemp)
    # 更新学习率
    curr_lr = smooth_step(10, 40, 100, 150, epoch)
    update_lr(optimizer, curr_lr)
    # 保存模型和一些参数、指标
    saveModel(model, epoch, test_acc, train_loss)
    # 记录时间
    out_epoch = time.time()
    print(f"use {(out_epoch-in_epoch)//60}min{(out_epoch-in_epoch)%60}s")
#$$
# In[1] 测试训练集
test(model, train_loader)


在这里插入代码片cuda
/home/megstudio/workspace/__20210502/_W28_04_BJGB3_BLN_S_C100
---  There is this folder!  ---
---  Create a new model...  ---
Epoch [1/200], Step [100/500] Loss: 4.5251
Epoch [1/200], Step [200/500] Loss: 4.5332
Epoch [1/200], Step [300/500] Loss: 4.4155
Epoch [1/200], Step [400/500] Loss: 4.4589
Epoch [1/200], Step [500/500] Loss: 4.4348
Accuracy of the model on the test images: 2.95 %
use 0.0min21.313775062561035s
Epoch [2/200], Step [100/500] Loss: 4.2417
Epoch [2/200], Step [200/500] Loss: 4.1760
Epoch [2/200], Step [300/500] Loss: 4.3673
Epoch [2/200], Step [400/500] Loss: 4.1980
Epoch [2/200], Step [500/500] Loss: 4.3332
Accuracy of the model on the test images: 8.77 %
use 0.0min20.538337469100952s
Epoch [3/200], Step [100/500] Loss: 4.1575
Epoch [3/200], Step [200/500] Loss: 3.9547
Epoch [3/200], Step [300/500] Loss: 4.3824
Epoch [3/200], Step [400/500] Loss: 4.2851
Epoch [3/200], Step [500/500] Loss: 4.3145
Accuracy of the model on the test images: 11.87 %
use 0.0min20.534307718276978s
Epoch [4/200], Step [100/500] Loss: 4.2549
Epoch [4/200], Step [200/500] Loss: 4.1143
Epoch [4/200], Step [300/500] Loss: 4.0363
Epoch [4/200], Step [400/500] Loss: 3.9180
Epoch [4/200], Step [500/500] Loss: 4.0309
Accuracy of the model on the test images: 13.54 %
use 0.0min20.53101944923401s
Epoch [5/200], Step [100/500] Loss: 4.3400
Epoch [5/200], Step [200/500] Loss: 4.1714
Epoch [5/200], Step [300/500] Loss: 3.7899
Epoch [5/200], Step [400/500] Loss: 4.0602
Epoch [5/200], Step [500/500] Loss: 3.6776
Accuracy of the model on the test images: 17.67 %
use 0.0min20.553125143051147s
Epoch [6/200], Step [100/500] Loss: 4.0879
Epoch [6/200], Step [200/500] Loss: 3.7493
Epoch [6/200], Step [300/500] Loss: 3.8425
Epoch [6/200], Step [400/500] Loss: 3.9041
Epoch [6/200], Step [500/500] Loss: 3.5192
Accuracy of the model on the test images: 17.4 %
use 0.0min20.53827691078186s
Epoch [7/200], Step [100/500] Loss: 3.3641
Epoch [7/200], Step [200/500] Loss: 3.4089
Epoch [7/200], Step [300/500] Loss: 3.6816
Epoch [7/200], Step [400/500] Loss: 4.2153
Epoch [7/200], Step [500/500] Loss: 4.0603
Accuracy of the model on the test images: 20.82 %
use 0.0min20.587323427200317s
Epoch [8/200], Step [100/500] Loss: 3.8739
Epoch [8/200], Step [200/500] Loss: 3.7661
Epoch [8/200], Step [300/500] Loss: 3.4724
Epoch [8/200], Step [400/500] Loss: 3.8320
Epoch [8/200], Step [500/500] Loss: 3.8333
Accuracy of the model on the test images: 22.37 %
use 0.0min20.577714204788208s
Epoch [9/200], Step [100/500] Loss: 3.7074
Epoch [9/200], Step [200/500] Loss: 3.8754
Epoch [9/200], Step [300/500] Loss: 3.2502
Epoch [9/200], Step [400/500] Loss: 3.1254
Epoch [9/200], Step [500/500] Loss: 3.5686
Accuracy of the model on the test images: 25.91 %
use 0.0min20.565754652023315s
Epoch [10/200], Step [100/500] Loss: 3.9443
Epoch [10/200], Step [200/500] Loss: 3.9528
Epoch [10/200], Step [300/500] Loss: 3.4973
Epoch [10/200], Step [400/500] Loss: 3.3094
Epoch [10/200], Step [500/500] Loss: 3.8865
Accuracy of the model on the test images: 28.39 %
use 0.0min20.67314100265503s
Epoch [11/200], Step [100/500] Loss: 4.1321
Epoch [11/200], Step [200/500] Loss: 3.0201
Epoch [11/200], Step [300/500] Loss: 2.9262
Epoch [11/200], Step [400/500] Loss: 3.8766
Epoch [11/200], Step [500/500] Loss: 3.8025
Accuracy of the model on the test images: 30.39 %
use 0.0min20.605696201324463s
Epoch [12/200], Step [100/500] Loss: 3.9981
Epoch [12/200], Step [200/500] Loss: 3.2076
Epoch [12/200], Step [300/500] Loss: 3.7684
Epoch [12/200], Step [400/500] Loss: 3.7398
Epoch [12/200], Step [500/500] Loss: 3.8506
Accuracy of the model on the test images: 32.4 %
use 0.0min20.603549003601074s
Epoch [13/200], Step [100/500] Loss: 3.2969
Epoch [13/200], Step [200/500] Loss: 3.7856
Epoch [13/200], Step [300/500] Loss: 2.8820
Epoch [13/200], Step [400/500] Loss: 3.9198
Epoch [13/200], Step [500/500] Loss: 3.6732
Accuracy of the model on the test images: 34.36 %
use 0.0min20.573402881622314s
Epoch [14/200], Step [100/500] Loss: 3.9373
Epoch [14/200], Step [200/500] Loss: 4.0874
Epoch [14/200], Step [300/500] Loss: 3.4495
Epoch [14/200], Step [400/500] Loss: 3.7622
Epoch [14/200], Step [500/500] Loss: 3.9038
Accuracy of the model on the test images: 34.08 %
use 0.0min20.558907747268677s
Epoch [15/200], Step [100/500] Loss: 3.7387
Epoch [15/200], Step [200/500] Loss: 3.8316
Epoch [15/200], Step [300/500] Loss: 3.6422
Epoch [15/200], Step [400/500] Loss: 2.9413
Epoch [15/200], Step [500/500] Loss: 3.2966
Accuracy of the model on the test images: 36.95 %
use 0.0min20.563921213150024s
Epoch [16/200], Step [100/500] Loss: 3.0023
Epoch [16/200], Step [200/500] Loss: 3.6259
Epoch [16/200], Step [300/500] Loss: 3.7841
Epoch [16/200], Step [400/500] Loss: 3.0014
Epoch [16/200], Step [500/500] Loss: 3.8931
Accuracy of the model on the test images: 38.75 %
use 0.0min20.559727907180786s
Epoch [17/200], Step [100/500] Loss: 3.7381
Epoch [17/200], Step [200/500] Loss: 2.4827
Epoch [17/200], Step [300/500] Loss: 3.4061
Epoch [17/200], Step [400/500] Loss: 3.5481
Epoch [17/200], Step [500/500] Loss: 3.6850
Accuracy of the model on the test images: 40.53 %
use 0.0min20.605258226394653s
Epoch [18/200], Step [100/500] Loss: 2.1450
Epoch [18/200], Step [200/500] Loss: 3.5551
Epoch [18/200], Step [300/500] Loss: 3.6963
Epoch [18/200], Step [400/500] Loss: 3.3295
Epoch [18/200], Step [500/500] Loss: 3.6757
Accuracy of the model on the test images: 43.47 %
use 0.0min20.565447092056274s
Epoch [19/200], Step [100/500] Loss: 3.4670
Epoch [19/200], Step [200/500] Loss: 2.3863
Epoch [19/200], Step [300/500] Loss: 2.0803
Epoch [19/200], Step [400/500] Loss: 2.9740
Epoch [19/200], Step [500/500] Loss: 2.6086
Accuracy of the model on the test images: 44.38 %
use 0.0min20.576362371444702s
Epoch [20/200], Step [100/500] Loss: 3.6958
Epoch [20/200], Step [200/500] Loss: 2.9528
Epoch [20/200], Step [300/500] Loss: 3.2621
Epoch [20/200], Step [400/500] Loss: 3.5945
Epoch [20/200], Step [500/500] Loss: 3.0555
Accuracy of the model on the test images: 45.25 %
use 0.0min20.55819010734558s
Epoch [21/200], Step [100/500] Loss: 2.7764
Epoch [21/200], Step [200/500] Loss: 2.7665
Epoch [21/200], Step [300/500] Loss: 3.2122
Epoch [21/200], Step [400/500] Loss: 3.3697
Epoch [21/200], Step [500/500] Loss: 3.6805
Accuracy of the model on the test images: 46.15 %
use 0.0min20.55286169052124s
Epoch [22/200], Step [100/500] Loss: 1.9792
Epoch [22/200], Step [200/500] Loss: 3.2696
Epoch [22/200], Step [300/500] Loss: 3.5111
Epoch [22/200], Step [400/500] Loss: 2.5218
Epoch [22/200], Step [500/500] Loss: 3.4128
Accuracy of the model on the test images: 50.65 %
use 0.0min20.55747890472412s
Epoch [23/200], Step [100/500] Loss: 2.9744
Epoch [23/200], Step [200/500] Loss: 2.2325
Epoch [23/200], Step [300/500] Loss: 3.1913
Epoch [23/200], Step [400/500] Loss: 2.3212
Epoch [23/200], Step [500/500] Loss: 3.4201
Accuracy of the model on the test images: 52.31 %
use 0.0min20.56867814064026s
Epoch [24/200], Step [100/500] Loss: 3.2548
Epoch [24/200], Step [200/500] Loss: 2.8263
Epoch [24/200], Step [300/500] Loss: 1.7920
Epoch [24/200], Step [400/500] Loss: 2.4103
Epoch [24/200], Step [500/500] Loss: 3.1759
Accuracy of the model on the test images: 52.24 %
use 0.0min20.573205947875977s
Epoch [25/200], Step [100/500] Loss: 3.2261
Epoch [25/200], Step [200/500] Loss: 1.7828
Epoch [25/200], Step [300/500] Loss: 3.4207
Epoch [25/200], Step [400/500] Loss: 2.0670
Epoch [25/200], Step [500/500] Loss: 2.9430
Accuracy of the model on the test images: 53.53 %
use 0.0min20.562430381774902s
Epoch [26/200], Step [100/500] Loss: 2.6081
Epoch [26/200], Step [200/500] Loss: 3.2208
Epoch [26/200], Step [300/500] Loss: 1.8349
Epoch [26/200], Step [400/500] Loss: 2.9786
Epoch [26/200], Step [500/500] Loss: 3.3320
Accuracy of the model on the test images: 54.59 %
use 0.0min20.57682728767395s
Epoch [27/200], Step [100/500] Loss: 3.0493
Epoch [27/200], Step [200/500] Loss: 1.6043
Epoch [27/200], Step [300/500] Loss: 2.5076
Epoch [27/200], Step [400/500] Loss: 3.2175
Epoch [27/200], Step [500/500] Loss: 3.2086
Accuracy of the model on the test images: 54.74 %
use 0.0min20.57382106781006s
Epoch [28/200], Step [100/500] Loss: 3.2156
Epoch [28/200], Step [200/500] Loss: 3.0681
Epoch [28/200], Step [300/500] Loss: 3.3450
Epoch [28/200], Step [400/500] Loss: 2.6080
Epoch [28/200], Step [500/500] Loss: 2.8209
Accuracy of the model on the test images: 56.12 %
use 0.0min20.601258039474487s
Epoch [29/200], Step [100/500] Loss: 1.3415
Epoch [29/200], Step [200/500] Loss: 2.2698
Epoch [29/200], Step [300/500] Loss: 2.0713
Epoch [29/200], Step [400/500] Loss: 3.1718
Epoch [29/200], Step [500/500] Loss: 2.4897
Accuracy of the model on the test images: 58.77 %
use 0.0min20.558830738067627s
Epoch [30/200], Step [100/500] Loss: 2.7607
Epoch [30/200], Step [200/500] Loss: 1.5064
Epoch [30/200], Step [300/500] Loss: 3.0134
Epoch [30/200], Step [400/500] Loss: 2.0962
Epoch [30/200], Step [500/500] Loss: 2.9495
Accuracy of the model on the test images: 58.98 %
use 0.0min20.562129259109497s
Epoch [31/200], Step [100/500] Loss: 3.0823
Epoch [31/200], Step [200/500] Loss: 2.8494
Epoch [31/200], Step [300/500] Loss: 2.3089
Epoch [31/200], Step [400/500] Loss: 1.9993
Epoch [31/200], Step [500/500] Loss: 2.2717
Accuracy of the model on the test images: 58.5 %
use 0.0min20.540247201919556s
Epoch [32/200], Step [100/500] Loss: 2.8952
Epoch [32/200], Step [200/500] Loss: 1.8841
Epoch [32/200], Step [300/500] Loss: 3.1238
Epoch [32/200], Step [400/500] Loss: 2.2653
Epoch [32/200], Step [500/500] Loss: 2.7214
Accuracy of the model on the test images: 60.02 %
use 0.0min20.57820773124695s
Epoch [33/200], Step [100/500] Loss: 2.5430
Epoch [33/200], Step [200/500] Loss: 3.0180
Epoch [33/200], Step [300/500] Loss: 2.8375
Epoch [33/200], Step [400/500] Loss: 3.1144
Epoch [33/200], Step [500/500] Loss: 1.3336
Accuracy of the model on the test images: 60.01 %
use 0.0min20.55961036682129s
Epoch [34/200], Step [100/500] Loss: 3.0738
Epoch [34/200], Step [200/500] Loss: 1.6780
Epoch [34/200], Step [300/500] Loss: 2.9272
Epoch [34/200], Step [400/500] Loss: 3.2433
Epoch [34/200], Step [500/500] Loss: 1.8242
Accuracy of the model on the test images: 61.06 %
use 0.0min20.55697011947632s
Epoch [35/200], Step [100/500] Loss: 2.7403
Epoch [35/200], Step [200/500] Loss: 2.8067
Epoch [35/200], Step [300/500] Loss: 3.2316
Epoch [35/200], Step [400/500] Loss: 2.7273
Epoch [35/200], Step [500/500] Loss: 2.9812
Accuracy of the model on the test images: 61.07 %
use 0.0min20.56986403465271s
Epoch [36/200], Step [100/500] Loss: 2.2424
Epoch [36/200], Step [200/500] Loss: 1.5960
Epoch [36/200], Step [300/500] Loss: 2.4490
Epoch [36/200], Step [400/500] Loss: 1.9833
Epoch [36/200], Step [500/500] Loss: 1.7756
Accuracy of the model on the test images: 61.99 %
use 0.0min20.587810516357422s
Epoch [37/200], Step [100/500] Loss: 2.3948
Epoch [37/200], Step [200/500] Loss: 2.7998
Epoch [37/200], Step [300/500] Loss: 2.7647
Epoch [37/200], Step [400/500] Loss: 2.5772
Epoch [37/200], Step [500/500] Loss: 2.8614
Accuracy of the model on the test images: 62.01 %
use 0.0min20.54871106147766s
Epoch [38/200], Step [100/500] Loss: 2.6874
Epoch [38/200], Step [200/500] Loss: 1.2753
Epoch [38/200], Step [300/500] Loss: 2.9217
Epoch [38/200], Step [400/500] Loss: 3.0492
Epoch [38/200], Step [500/500] Loss: 2.7927
Accuracy of the model on the test images: 61.13 %
use 0.0min20.576314449310303s
Epoch [39/200], Step [100/500] Loss: 2.7591
Epoch [39/200], Step [200/500] Loss: 2.4770
Epoch [39/200], Step [300/500] Loss: 2.4154
Epoch [39/200], Step [400/500] Loss: 2.2822
Epoch [39/200], Step [500/500] Loss: 1.4424
Accuracy of the model on the test images: 63.91 %
use 0.0min20.582370042800903s
Epoch [40/200], Step [100/500] Loss: 2.9856
Epoch [40/200], Step [200/500] Loss: 2.6278
Epoch [40/200], Step [300/500] Loss: 2.9802
Epoch [40/200], Step [400/500] Loss: 1.7501
Epoch [40/200], Step [500/500] Loss: 2.7796
Accuracy of the model on the test images: 63.5 %
use 0.0min20.566586017608643s
Epoch [41/200], Step [100/500] Loss: 2.8127
Epoch [41/200], Step [200/500] Loss: 1.0172
Epoch [41/200], Step [300/500] Loss: 2.7400
Epoch [41/200], Step [400/500] Loss: 2.6278
Epoch [41/200], Step [500/500] Loss: 3.0157
Accuracy of the model on the test images: 63.42 %
use 0.0min20.55940055847168s
Epoch [42/200], Step [100/500] Loss: 1.1076
Epoch [42/200], Step [200/500] Loss: 2.9837
Epoch [42/200], Step [300/500] Loss: 2.6981
Epoch [42/200], Step [400/500] Loss: 2.1914
Epoch [42/200], Step [500/500] Loss: 1.8929
Accuracy of the model on the test images: 61.08 %
use 0.0min20.563587427139282s
Epoch [43/200], Step [100/500] Loss: 2.7030
Epoch [43/200], Step [200/500] Loss: 2.6583
Epoch [43/200], Step [300/500] Loss: 1.3961
Epoch [43/200], Step [400/500] Loss: 2.4727
Epoch [43/200], Step [500/500] Loss: 1.6471
Accuracy of the model on the test images: 63.71 %
use 0.0min20.584805250167847s
Epoch [44/200], Step [100/500] Loss: 2.6830
Epoch [44/200], Step [200/500] Loss: 3.0156
Epoch [44/200], Step [300/500] Loss: 2.1286
Epoch [44/200], Step [400/500] Loss: 2.8233
Epoch [44/200], Step [500/500] Loss: 1.2222
Accuracy of the model on the test images: 64.43 %
use 0.0min20.55425238609314s
Epoch [45/200], Step [100/500] Loss: 2.6584
Epoch [45/200], Step [200/500] Loss: 2.6890
Epoch [45/200], Step [300/500] Loss: 3.1283
Epoch [45/200], Step [400/500] Loss: 2.6737
Epoch [45/200], Step [500/500] Loss: 2.1185
Accuracy of the model on the test images: 65.41 %
use 0.0min20.57489562034607s
Epoch [46/200], Step [100/500] Loss: 2.8614
Epoch [46/200], Step [200/500] Loss: 2.8155
Epoch [46/200], Step [300/500] Loss: 2.8558
Epoch [46/200], Step [400/500] Loss: 2.4669
Epoch [46/200], Step [500/500] Loss: 2.6695
Accuracy of the model on the test images: 65.98 %
use 0.0min20.580721616744995s
Epoch [47/200], Step [100/500] Loss: 2.1174
Epoch [47/200], Step [200/500] Loss: 0.9854
Epoch [47/200], Step [300/500] Loss: 2.1330
Epoch [47/200], Step [400/500] Loss: 2.4373
Epoch [47/200], Step [500/500] Loss: 2.5382
Accuracy of the model on the test images: 66.56 %
use 0.0min20.548205137252808s
Epoch [48/200], Step [100/500] Loss: 2.8388
Epoch [48/200], Step [200/500] Loss: 2.7994
Epoch [48/200], Step [300/500] Loss: 1.3432
Epoch [48/200], Step [400/500] Loss: 2.0102
Epoch [48/200], Step [500/500] Loss: 2.7221
Accuracy of the model on the test images: 65.37 %
use 0.0min20.57545757293701s
Epoch [49/200], Step [100/500] Loss: 2.7333
Epoch [49/200], Step [200/500] Loss: 1.1249
Epoch [49/200], Step [300/500] Loss: 2.1527
Epoch [49/200], Step [400/500] Loss: 2.6832
Epoch [49/200], Step [500/500] Loss: 1.0853
Accuracy of the model on the test images: 67.23 %
use 0.0min20.54992938041687s
Epoch [50/200], Step [100/500] Loss: 2.1665
Epoch [50/200], Step [200/500] Loss: 1.8453
Epoch [50/200], Step [300/500] Loss: 2.5208
Epoch [50/200], Step [400/500] Loss: 2.0102
Epoch [50/200], Step [500/500] Loss: 1.5296
Accuracy of the model on the test images: 66.47 %
use 0.0min20.569560766220093s
Epoch [51/200], Step [100/500] Loss: 2.2455
Epoch [51/200], Step [200/500] Loss: 1.1350
Epoch [51/200], Step [300/500] Loss: 2.7719
Epoch [51/200], Step [400/500] Loss: 2.5876
Epoch [51/200], Step [500/500] Loss: 2.0808
Accuracy of the model on the test images: 68.07 %
use 0.0min20.59921383857727s
Epoch [52/200], Step [100/500] Loss: 0.9167
Epoch [52/200], Step [200/500] Loss: 2.4627
Epoch [52/200], Step [300/500] Loss: 2.3724
Epoch [52/200], Step [400/500] Loss: 0.7734
Epoch [52/200], Step [500/500] Loss: 2.0108
Accuracy of the model on the test images: 66.56 %
use 0.0min20.575576543807983s
Epoch [53/200], Step [100/500] Loss: 1.9589
Epoch [53/200], Step [200/500] Loss: 1.3028
Epoch [53/200], Step [300/500] Loss: 2.4040
Epoch [53/200], Step [400/500] Loss: 1.3875
Epoch [53/200], Step [500/500] Loss: 1.6074
Accuracy of the model on the test images: 66.79 %
use 0.0min20.563728094100952s
Epoch [54/200], Step [100/500] Loss: 2.7122
Epoch [54/200], Step [200/500] Loss: 1.9426
Epoch [54/200], Step [300/500] Loss: 2.4717
Epoch [54/200], Step [400/500] Loss: 1.2387
Epoch [54/200], Step [500/500] Loss: 0.9918
Accuracy of the model on the test images: 67.79 %
use 0.0min20.57853603363037s
Epoch [55/200], Step [100/500] Loss: 2.4608
Epoch [55/200], Step [200/500] Loss: 2.3345
Epoch [55/200], Step [300/500] Loss: 2.4966
Epoch [55/200], Step [400/500] Loss: 2.3704
Epoch [55/200], Step [500/500] Loss: 1.0572
Accuracy of the model on the test images: 66.99 %
use 0.0min20.571059703826904s
Epoch [56/200], Step [100/500] Loss: 1.4127
Epoch [56/200], Step [200/500] Loss: 0.9468
Epoch [56/200], Step [300/500] Loss: 2.7228
Epoch [56/200], Step [400/500] Loss: 0.9251
Epoch [56/200], Step [500/500] Loss: 1.7493
Accuracy of the model on the test images: 68.11 %
use 0.0min20.559414625167847s
Epoch [57/200], Step [100/500] Loss: 2.4703
Epoch [57/200], Step [200/500] Loss: 2.6340
Epoch [57/200], Step [300/500] Loss: 2.6556
Epoch [57/200], Step [400/500] Loss: 1.9797
Epoch [57/200], Step [500/500] Loss: 1.7072
Accuracy of the model on the test images: 68.33 %
use 0.0min20.571043491363525s
Epoch [58/200], Step [100/500] Loss: 2.0545
Epoch [58/200], Step [200/500] Loss: 1.9410
Epoch [58/200], Step [300/500] Loss: 2.4254
Epoch [58/200], Step [400/500] Loss: 2.3523
Epoch [58/200], Step [500/500] Loss: 2.0542
Accuracy of the model on the test images: 67.36 %
use 0.0min20.5789737701416s
Epoch [59/200], Step [100/500] Loss: 2.1255
Epoch [59/200], Step [200/500] Loss: 2.2644
Epoch [59/200], Step [300/500] Loss: 1.5045
Epoch [59/200], Step [400/500] Loss: 2.9253
Epoch [59/200], Step [500/500] Loss: 2.2406
Accuracy of the model on the test images: 67.94 %
use 0.0min20.57670783996582s
Epoch [60/200], Step [100/500] Loss: 1.9240
Epoch [60/200], Step [200/500] Loss: 2.4129
Epoch [60/200], Step [300/500] Loss: 2.3032
Epoch [60/200], Step [400/500] Loss: 2.2535
Epoch [60/200], Step [500/500] Loss: 1.8229
Accuracy of the model on the test images: 69.48 %
use 0.0min20.565730571746826s
Epoch [61/200], Step [100/500] Loss: 0.5190
Epoch [61/200], Step [200/500] Loss: 2.0898
Epoch [61/200], Step [300/500] Loss: 1.0934
Epoch [61/200], Step [400/500] Loss: 1.7987
Epoch [61/200], Step [500/500] Loss: 1.0228
Accuracy of the model on the test images: 67.83 %
use 0.0min20.598429441452026s
Epoch [62/200], Step [100/500] Loss: 2.7897
Epoch [62/200], Step [200/500] Loss: 1.4987
Epoch [62/200], Step [300/500] Loss: 1.8081
Epoch [62/200], Step [400/500] Loss: 2.5582
Epoch [62/200], Step [500/500] Loss: 2.6745
Accuracy of the model on the test images: 68.44 %
use 0.0min20.624147176742554s
Epoch [63/200], Step [100/500] Loss: 2.3224
Epoch [63/200], Step [200/500] Loss: 2.5201
Epoch [63/200], Step [300/500] Loss: 0.5925
Epoch [63/200], Step [400/500] Loss: 1.0193
Epoch [63/200], Step [500/500] Loss: 0.8275
Accuracy of the model on the test images: 69.22 %
use 0.0min20.572014331817627s
Epoch [64/200], Step [100/500] Loss: 1.8970
Epoch [64/200], Step [200/500] Loss: 2.0111
Epoch [64/200], Step [300/500] Loss: 2.3616
Epoch [64/200], Step [400/500] Loss: 2.1698
Epoch [64/200], Step [500/500] Loss: 1.5965
Accuracy of the model on the test images: 68.63 %
use 0.0min20.575780868530273s
Epoch [65/200], Step [100/500] Loss: 1.5909
Epoch [65/200], Step [200/500] Loss: 1.4618
Epoch [65/200], Step [300/500] Loss: 2.5006
Epoch [65/200], Step [400/500] Loss: 0.6584
Epoch [65/200], Step [500/500] Loss: 1.5347
Accuracy of the model on the test images: 69.31 %
use 0.0min20.563440322875977s
Epoch [66/200], Step [100/500] Loss: 2.4060
Epoch [66/200], Step [200/500] Loss: 2.6861
Epoch [66/200], Step [300/500] Loss: 2.1461
Epoch [66/200], Step [400/500] Loss: 1.9965
Epoch [66/200], Step [500/500] Loss: 1.9670
Accuracy of the model on the test images: 69.66 %
use 0.0min20.569555521011353s
Epoch [67/200], Step [100/500] Loss: 0.4118
Epoch [67/200], Step [200/500] Loss: 2.5139
Epoch [67/200], Step [300/500] Loss: 2.5679
Epoch [67/200], Step [400/500] Loss: 2.4008
Epoch [67/200], Step [500/500] Loss: 2.0521
Accuracy of the model on the test images: 69.69 %
use 0.0min20.61587929725647s
Epoch [68/200], Step [100/500] Loss: 1.6595
Epoch [68/200], Step [200/500] Loss: 2.3468
Epoch [68/200], Step [300/500] Loss: 2.4786
Epoch [68/200], Step [400/500] Loss: 2.5231
Epoch [68/200], Step [500/500] Loss: 2.3521
Accuracy of the model on the test images: 69.94 %
use 0.0min20.554180145263672s
Epoch [69/200], Step [100/500] Loss: 2.4418
Epoch [69/200], Step [200/500] Loss: 2.6581
Epoch [69/200], Step [300/500] Loss: 2.1739
Epoch [69/200], Step [400/500] Loss: 2.0204
Epoch [69/200], Step [500/500] Loss: 2.5698
Accuracy of the model on the test images: 69.53 %
use 0.0min20.614558458328247s
Epoch [70/200], Step [100/500] Loss: 2.2597
Epoch [70/200], Step [200/500] Loss: 1.5124
Epoch [70/200], Step [300/500] Loss: 1.6330
Epoch [70/200], Step [400/500] Loss: 0.7440
Epoch [70/200], Step [500/500] Loss: 2.3981
Accuracy of the model on the test images: 70.21 %
use 0.0min20.55963921546936s
Epoch [71/200], Step [100/500] Loss: 1.9647
Epoch [71/200], Step [200/500] Loss: 1.0061
Epoch [71/200], Step [300/500] Loss: 1.6910
Epoch [71/200], Step [400/500] Loss: 2.4693
Epoch [71/200], Step [500/500] Loss: 2.3597
Accuracy of the model on the test images: 69.73 %
use 0.0min20.597806215286255s
Epoch [72/200], Step [100/500] Loss: 1.2391
Epoch [72/200], Step [200/500] Loss: 1.7451
Epoch [72/200], Step [300/500] Loss: 1.4576
Epoch [72/200], Step [400/500] Loss: 2.6661
Epoch [72/200], Step [500/500] Loss: 1.8903
Accuracy of the model on the test images: 70.11 %
use 0.0min20.533710479736328s
Epoch [73/200], Step [100/500] Loss: 2.4056
Epoch [73/200], Step [200/500] Loss: 2.2574
Epoch [73/200], Step [300/500] Loss: 2.4138
Epoch [73/200], Step [400/500] Loss: 2.5121
Epoch [73/200], Step [500/500] Loss: 2.6018
Accuracy of the model on the test images: 69.59 %
use 0.0min20.558839797973633s
Epoch [74/200], Step [100/500] Loss: 1.2735
Epoch [74/200], Step [200/500] Loss: 2.2442
Epoch [74/200], Step [300/500] Loss: 2.3996
Epoch [74/200], Step [400/500] Loss: 2.0632
Epoch [74/200], Step [500/500] Loss: 2.4572
Accuracy of the model on the test images: 69.71 %
use 0.0min20.57104754447937s
Epoch [75/200], Step [100/500] Loss: 2.4749
Epoch [75/200], Step [200/500] Loss: 2.3034
Epoch [75/200], Step [300/500] Loss: 2.7674
Epoch [75/200], Step [400/500] Loss: 2.3820
Epoch [75/200], Step [500/500] Loss: 2.1781
Accuracy of the model on the test images: 69.22 %
use 0.0min20.55558466911316s
Epoch [76/200], Step [100/500] Loss: 0.5393
Epoch [76/200], Step [200/500] Loss: 2.1347
Epoch [76/200], Step [300/500] Loss: 2.0176
Epoch [76/200], Step [400/500] Loss: 2.3896
Epoch [76/200], Step [500/500] Loss: 2.1903
Accuracy of the model on the test images: 69.99 %
use 0.0min20.568068265914917s
Epoch [77/200], Step [100/500] Loss: 2.2278
Epoch [77/200], Step [200/500] Loss: 1.4643
Epoch [77/200], Step [300/500] Loss: 2.5015
Epoch [77/200], Step [400/500] Loss: 2.1213
Epoch [77/200], Step [500/500] Loss: 1.3269
Accuracy of the model on the test images: 69.77 %
use 0.0min20.610535860061646s
Epoch [78/200], Step [100/500] Loss: 2.3664
Epoch [78/200], Step [200/500] Loss: 2.0794
Epoch [78/200], Step [300/500] Loss: 1.6979
Epoch [78/200], Step [400/500] Loss: 1.7097
Epoch [78/200], Step [500/500] Loss: 1.3437
Accuracy of the model on the test images: 70.81 %
use 0.0min20.569110870361328s
Epoch [79/200], Step [100/500] Loss: 2.1547
Epoch [79/200], Step [200/500] Loss: 1.7811
Epoch [79/200], Step [300/500] Loss: 2.1639
Epoch [79/200], Step [400/500] Loss: 1.5924
Epoch [79/200], Step [500/500] Loss: 0.7399
Accuracy of the model on the test images: 70.25 %
use 0.0min20.58094096183777s
Epoch [80/200], Step [100/500] Loss: 2.1095
Epoch [80/200], Step [200/500] Loss: 1.3439
Epoch [80/200], Step [300/500] Loss: 2.5862
Epoch [80/200], Step [400/500] Loss: 2.3377
Epoch [80/200], Step [500/500] Loss: 2.2636
Accuracy of the model on the test images: 70.62 %
use 0.0min20.559268712997437s
Epoch [81/200], Step [100/500] Loss: 2.5232
Epoch [81/200], Step [200/500] Loss: 1.9517
Epoch [81/200], Step [300/500] Loss: 2.0707
Epoch [81/200], Step [400/500] Loss: 2.2953
Epoch [81/200], Step [500/500] Loss: 0.6598
Accuracy of the model on the test images: 70.1 %
use 0.0min20.59356713294983s
Epoch [82/200], Step [100/500] Loss: 2.0120
Epoch [82/200], Step [200/500] Loss: 1.9701
Epoch [82/200], Step [300/500] Loss: 2.2314
Epoch [82/200], Step [400/500] Loss: 2.1266
Epoch [82/200], Step [500/500] Loss: 1.5196
Accuracy of the model on the test images: 70.65 %
use 0.0min20.570524215698242s
Epoch [83/200], Step [100/500] Loss: 1.4251
Epoch [83/200], Step [200/500] Loss: 2.2970
Epoch [83/200], Step [300/500] Loss: 2.6220
Epoch [83/200], Step [400/500] Loss: 2.2916
Epoch [83/200], Step [500/500] Loss: 2.4677
Accuracy of the model on the test images: 70.83 %
use 0.0min20.57495927810669s
Epoch [84/200], Step [100/500] Loss: 1.1092
Epoch [84/200], Step [200/500] Loss: 2.5165
Epoch [84/200], Step [300/500] Loss: 0.7905
Epoch [84/200], Step [400/500] Loss: 1.8410
Epoch [84/200], Step [500/500] Loss: 2.3548
Accuracy of the model on the test images: 70.48 %
use 0.0min20.583019256591797s
Epoch [85/200], Step [100/500] Loss: 2.4547
Epoch [85/200], Step [200/500] Loss: 1.9760
Epoch [85/200], Step [300/500] Loss: 2.2574
Epoch [85/200], Step [400/500] Loss: 1.4774
Epoch [85/200], Step [500/500] Loss: 2.0934
Accuracy of the model on the test images: 71.42 %
use 0.0min20.60036563873291s
Epoch [86/200], Step [100/500] Loss: 2.0185
Epoch [86/200], Step [200/500] Loss: 2.1625
Epoch [86/200], Step [300/500] Loss: 2.5006
Epoch [86/200], Step [400/500] Loss: 2.4513
Epoch [86/200], Step [500/500] Loss: 2.0119
Accuracy of the model on the test images: 70.49 %
use 0.0min20.592283964157104s
Epoch [87/200], Step [100/500] Loss: 1.5303
Epoch [87/200], Step [200/500] Loss: 2.1887
Epoch [87/200], Step [300/500] Loss: 2.3164
Epoch [87/200], Step [400/500] Loss: 2.4048
Epoch [87/200], Step [500/500] Loss: 2.3217
Accuracy of the model on the test images: 70.14 %
use 0.0min20.558128356933594s
Epoch [88/200], Step [100/500] Loss: 0.7063
Epoch [88/200], Step [200/500] Loss: 1.0208
Epoch [88/200], Step [300/500] Loss: 2.1137
Epoch [88/200], Step [400/500] Loss: 1.9834
Epoch [88/200], Step [500/500] Loss: 1.6025
Accuracy of the model on the test images: 70.52 %
use 0.0min20.59108805656433s
Epoch [89/200], Step [100/500] Loss: 0.6478
Epoch [89/200], Step [200/500] Loss: 2.1763
Epoch [89/200], Step [300/500] Loss: 1.8871
Epoch [89/200], Step [400/500] Loss: 2.1027
Epoch [89/200], Step [500/500] Loss: 1.1530
Accuracy of the model on the test images: 70.89 %
use 0.0min20.575583934783936s
Epoch [90/200], Step [100/500] Loss: 0.7405
Epoch [90/200], Step [200/500] Loss: 1.6733
Epoch [90/200], Step [300/500] Loss: 1.0202
Epoch [90/200], Step [400/500] Loss: 0.5583
Epoch [90/200], Step [500/500] Loss: 2.0358
Accuracy of the model on the test images: 70.96 %
use 0.0min20.558996438980103s
Epoch [91/200], Step [100/500] Loss: 1.1312
Epoch [91/200], Step [200/500] Loss: 1.4149
Epoch [91/200], Step [300/500] Loss: 2.5919
Epoch [91/200], Step [400/500] Loss: 2.0737
Epoch [91/200], Step [500/500] Loss: 2.2732
Accuracy of the model on the test images: 70.94 %
use 0.0min20.59275484085083s
Epoch [92/200], Step [100/500] Loss: 2.0307
Epoch [92/200], Step [200/500] Loss: 1.9069
Epoch [92/200], Step [300/500] Loss: 2.4373
Epoch [92/200], Step [400/500] Loss: 2.2249
Epoch [92/200], Step [500/500] Loss: 1.7237
Accuracy of the model on the test images: 69.91 %
use 0.0min20.58454942703247s
Epoch [93/200], Step [100/500] Loss: 1.2309
Epoch [93/200], Step [200/500] Loss: 1.6310
Epoch [93/200], Step [300/500] Loss: 1.1844
Epoch [93/200], Step [400/500] Loss: 2.4646
Epoch [93/200], Step [500/500] Loss: 1.0722
Accuracy of the model on the test images: 70.68 %
use 0.0min20.585512161254883s
Epoch [94/200], Step [100/500] Loss: 1.0893
Epoch [94/200], Step [200/500] Loss: 2.1244
Epoch [94/200], Step [300/500] Loss: 2.3294
Epoch [94/200], Step [400/500] Loss: 1.9373
Epoch [94/200], Step [500/500] Loss: 1.7801
Accuracy of the model on the test images: 70.46 %
use 0.0min20.566587686538696s
Epoch [95/200], Step [100/500] Loss: 1.9155
Epoch [95/200], Step [200/500] Loss: 2.0733
Epoch [95/200], Step [300/500] Loss: 1.2881
Epoch [95/200], Step [400/500] Loss: 1.1430
Epoch [95/200], Step [500/500] Loss: 1.3174
Accuracy of the model on the test images: 71.38 %
use 0.0min20.579296588897705s
Epoch [96/200], Step [100/500] Loss: 2.3137
Epoch [96/200], Step [200/500] Loss: 2.0052
Epoch [96/200], Step [300/500] Loss: 1.7943
Epoch [96/200], Step [400/500] Loss: 2.1875
Epoch [96/200], Step [500/500] Loss: 2.0805
Accuracy of the model on the test images: 71.04 %
use 0.0min20.5848228931427s
Epoch [97/200], Step [100/500] Loss: 1.6475
Epoch [97/200], Step [200/500] Loss: 1.2999
Epoch [97/200], Step [300/500] Loss: 0.5330
Epoch [97/200], Step [400/500] Loss: 2.1006
Epoch [97/200], Step [500/500] Loss: 2.0848
Accuracy of the model on the test images: 71.5 %
use 0.0min20.578545331954956s
Epoch [98/200], Step [100/500] Loss: 1.8841
Epoch [98/200], Step [200/500] Loss: 0.8905
Epoch [98/200], Step [300/500] Loss: 2.1971
Epoch [98/200], Step [400/500] Loss: 0.3851
Epoch [98/200], Step [500/500] Loss: 1.8517
Accuracy of the model on the test images: 70.75 %
use 0.0min20.55072855949402s
Epoch [99/200], Step [100/500] Loss: 1.7956
Epoch [99/200], Step [200/500] Loss: 1.1191
Epoch [99/200], Step [300/500] Loss: 2.0882
Epoch [99/200], Step [400/500] Loss: 1.6622
Epoch [99/200], Step [500/500] Loss: 2.0943
Accuracy of the model on the test images: 71.13 %
use 0.0min20.59413242340088s
Epoch [100/200], Step [100/500] Loss: 0.3222
Epoch [100/200], Step [200/500] Loss: 1.0598
Epoch [100/200], Step [300/500] Loss: 2.1213
Epoch [100/200], Step [400/500] Loss: 2.1117
Epoch [100/200], Step [500/500] Loss: 1.9403
Accuracy of the model on the test images: 71.31 %
use 0.0min20.62356424331665s
Epoch [101/200], Step [100/500] Loss: 1.9459
Epoch [101/200], Step [200/500] Loss: 0.7791
Epoch [101/200], Step [300/500] Loss: 1.8812
Epoch [101/200], Step [400/500] Loss: 0.9668
Epoch [101/200], Step [500/500] Loss: 1.2563
Accuracy of the model on the test images: 71.06 %
use 0.0min20.577518224716187s
Epoch [102/200], Step [100/500] Loss: 1.4139
Epoch [102/200], Step [200/500] Loss: 2.0665
Epoch [102/200], Step [300/500] Loss: 2.1635
Epoch [102/200], Step [400/500] Loss: 2.5603
Epoch [102/200], Step [500/500] Loss: 1.3941
Accuracy of the model on the test images: 71.85 %
use 0.0min20.579555988311768s
Epoch [103/200], Step [100/500] Loss: 0.4678
Epoch [103/200], Step [200/500] Loss: 0.2486
Epoch [103/200], Step [300/500] Loss: 1.3987
Epoch [103/200], Step [400/500] Loss: 2.1079
Epoch [103/200], Step [500/500] Loss: 1.7808
Accuracy of the model on the test images: 74.55 %
use 0.0min20.559417963027954s
Epoch [104/200], Step [100/500] Loss: 0.7784
Epoch [104/200], Step [200/500] Loss: 2.0021
Epoch [104/200], Step [300/500] Loss: 1.6035
Epoch [104/200], Step [400/500] Loss: 1.7185
Epoch [104/200], Step [500/500] Loss: 0.7297
Accuracy of the model on the test images: 75.14 %
use 0.0min20.567699670791626s
Epoch [105/200], Step [100/500] Loss: 1.7225
Epoch [105/200], Step [200/500] Loss: 1.1658
Epoch [105/200], Step [300/500] Loss: 0.6898
Epoch [105/200], Step [400/500] Loss: 1.6498
Epoch [105/200], Step [500/500] Loss: 1.7643
Accuracy of the model on the test images: 74.82 %
use 0.0min20.575838565826416s
Epoch [106/200], Step [100/500] Loss: 1.2480
Epoch [106/200], Step [200/500] Loss: 1.7975
Epoch [106/200], Step [300/500] Loss: 1.5042
Epoch [106/200], Step [400/500] Loss: 1.6249
Epoch [106/200], Step [500/500] Loss: 0.9821
Accuracy of the model on the test images: 75.0 %
use 0.0min20.591930866241455s
Epoch [107/200], Step [100/500] Loss: 1.8657
Epoch [107/200], Step [200/500] Loss: 1.7952
Epoch [107/200], Step [300/500] Loss: 1.6980
Epoch [107/200], Step [400/500] Loss: 0.9312
Epoch [107/200], Step [500/500] Loss: 1.4100
Accuracy of the model on the test images: 74.87 %
use 0.0min20.5633327960968s
Epoch [108/200], Step [100/500] Loss: 1.7571
Epoch [108/200], Step [200/500] Loss: 1.0123
Epoch [108/200], Step [300/500] Loss: 1.5292
Epoch [108/200], Step [400/500] Loss: 1.9204
Epoch [108/200], Step [500/500] Loss: 1.6726
Accuracy of the model on the test images: 75.44 %
use 0.0min20.605122327804565s
Epoch [109/200], Step [100/500] Loss: 1.0595
Epoch [109/200], Step [200/500] Loss: 1.8087
Epoch [109/200], Step [300/500] Loss: 1.0939
Epoch [109/200], Step [400/500] Loss: 1.6037
Epoch [109/200], Step [500/500] Loss: 2.0924
Accuracy of the model on the test images: 75.4 %
use 0.0min20.572604179382324s
Epoch [110/200], Step [100/500] Loss: 2.0441
Epoch [110/200], Step [200/500] Loss: 1.1455
Epoch [110/200], Step [300/500] Loss: 0.7206
Epoch [110/200], Step [400/500] Loss: 0.6228
Epoch [110/200], Step [500/500] Loss: 1.8622
Accuracy of the model on the test images: 75.5 %
use 0.0min20.553505897521973s
Epoch [111/200], Step [100/500] Loss: 1.0353
Epoch [111/200], Step [200/500] Loss: 2.2427
Epoch [111/200], Step [300/500] Loss: 2.1725
Epoch [111/200], Step [400/500] Loss: 1.5280
Epoch [111/200], Step [500/500] Loss: 1.4664
Accuracy of the model on the test images: 75.29 %
use 0.0min20.563225269317627s
Epoch [112/200], Step [100/500] Loss: 1.1391
Epoch [112/200], Step [200/500] Loss: 0.4924
Epoch [112/200], Step [300/500] Loss: 0.2584
Epoch [112/200], Step [400/500] Loss: 1.6328
Epoch [112/200], Step [500/500] Loss: 0.6042
Accuracy of the model on the test images: 75.19 %
use 0.0min20.582221746444702s
Epoch [113/200], Step [100/500] Loss: 1.8356
Epoch [113/200], Step [200/500] Loss: 1.5393
Epoch [113/200], Step [300/500] Loss: 1.7863
Epoch [113/200], Step [400/500] Loss: 1.8563
Epoch [113/200], Step [500/500] Loss: 2.1537
Accuracy of the model on the test images: 75.4 %
use 0.0min20.558560371398926s
Epoch [114/200], Step [100/500] Loss: 1.2635
Epoch [114/200], Step [200/500] Loss: 0.7171
Epoch [114/200], Step [300/500] Loss: 1.8709
Epoch [114/200], Step [400/500] Loss: 2.0476
Epoch [114/200], Step [500/500] Loss: 1.8067
Accuracy of the model on the test images: 75.31 %
use 0.0min20.567386865615845s
Epoch [115/200], Step [100/500] Loss: 1.8186
Epoch [115/200], Step [200/500] Loss: 1.0045
Epoch [115/200], Step [300/500] Loss: 0.8481
Epoch [115/200], Step [400/500] Loss: 2.3573
Epoch [115/200], Step [500/500] Loss: 1.4470
Accuracy of the model on the test images: 75.49 %
use 0.0min20.611566066741943s
Epoch [116/200], Step [100/500] Loss: 2.1492
Epoch [116/200], Step [200/500] Loss: 1.5311
Epoch [116/200], Step [300/500] Loss: 1.7044
Epoch [116/200], Step [400/500] Loss: 2.0091
Epoch [116/200], Step [500/500] Loss: 1.5411
Accuracy of the model on the test images: 75.57 %
use 0.0min20.577484369277954s
Epoch [117/200], Step [100/500] Loss: 1.2268
Epoch [117/200], Step [200/500] Loss: 0.1760
Epoch [117/200], Step [300/500] Loss: 0.1852
Epoch [117/200], Step [400/500] Loss: 0.7704
Epoch [117/200], Step [500/500] Loss: 2.0371
Accuracy of the model on the test images: 75.53 %
use 0.0min20.580153703689575s
Epoch [118/200], Step [100/500] Loss: 1.7165
Epoch [118/200], Step [200/500] Loss: 1.3778
Epoch [118/200], Step [300/500] Loss: 1.2850
Epoch [118/200], Step [400/500] Loss: 1.8418
Epoch [118/200], Step [500/500] Loss: 0.8564
Accuracy of the model on the test images: 75.09 %
use 0.0min20.5632164478302s
Epoch [119/200], Step [100/500] Loss: 1.7683
Epoch [119/200], Step [200/500] Loss: 1.7948
Epoch [119/200], Step [300/500] Loss: 1.5932
Epoch [119/200], Step [400/500] Loss: 1.6570
Epoch [119/200], Step [500/500] Loss: 1.4463
Accuracy of the model on the test images: 75.54 %
use 0.0min20.575408935546875s
Epoch [120/200], Step [100/500] Loss: 1.2803
Epoch [120/200], Step [200/500] Loss: 1.9159
Epoch [120/200], Step [300/500] Loss: 1.8203
Epoch [120/200], Step [400/500] Loss: 1.4216
Epoch [120/200], Step [500/500] Loss: 1.4928
Accuracy of the model on the test images: 75.46 %
use 0.0min20.57673954963684s
Epoch [121/200], Step [100/500] Loss: 1.5456
Epoch [121/200], Step [200/500] Loss: 2.0293
Epoch [121/200], Step [300/500] Loss: 1.2598
Epoch [121/200], Step [400/500] Loss: 1.8437
Epoch [121/200], Step [500/500] Loss: 1.9020
Accuracy of the model on the test images: 75.83 %
use 0.0min20.5976459980011s
Epoch [122/200], Step [100/500] Loss: 2.3689
Epoch [122/200], Step [200/500] Loss: 2.0162
Epoch [122/200], Step [300/500] Loss: 1.6390
Epoch [122/200], Step [400/500] Loss: 1.5606
Epoch [122/200], Step [500/500] Loss: 1.8211
Accuracy of the model on the test images: 75.76 %
use 0.0min20.56261420249939s
Epoch [123/200], Step [100/500] Loss: 1.5309
Epoch [123/200], Step [200/500] Loss: 0.9289
Epoch [123/200], Step [300/500] Loss: 1.8214
Epoch [123/200], Step [400/500] Loss: 1.7606
Epoch [123/200], Step [500/500] Loss: 1.4932
Accuracy of the model on the test images: 75.46 %
use 0.0min20.581864833831787s
Epoch [124/200], Step [100/500] Loss: 1.3017
Epoch [124/200], Step [200/500] Loss: 1.5079
Epoch [124/200], Step [300/500] Loss: 1.9221
Epoch [124/200], Step [400/500] Loss: 1.9697
Epoch [124/200], Step [500/500] Loss: 0.9600
Accuracy of the model on the test images: 75.57 %
use 0.0min20.589545965194702s
Epoch [125/200], Step [100/500] Loss: 1.4066
Epoch [125/200], Step [200/500] Loss: 1.0608
Epoch [125/200], Step [300/500] Loss: 2.1675
Epoch [125/200], Step [400/500] Loss: 0.1017
Epoch [125/200], Step [500/500] Loss: 0.1973
Accuracy of the model on the test images: 75.44 %
use 0.0min20.59250807762146s
Epoch [126/200], Step [100/500] Loss: 1.9236
Epoch [126/200], Step [200/500] Loss: 0.7173
Epoch [126/200], Step [300/500] Loss: 1.6922
Epoch [126/200], Step [400/500] Loss: 0.2032
Epoch [126/200], Step [500/500] Loss: 1.0125
Accuracy of the model on the test images: 75.76 %
use 0.0min20.574411630630493s
Epoch [127/200], Step [100/500] Loss: 1.0648
Epoch [127/200], Step [200/500] Loss: 1.8413
Epoch [127/200], Step [300/500] Loss: 0.8166
Epoch [127/200], Step [400/500] Loss: 0.7186
Epoch [127/200], Step [500/500] Loss: 1.9249
Accuracy of the model on the test images: 75.74 %
use 0.0min20.576762437820435s
Epoch [128/200], Step [100/500] Loss: 1.3528
Epoch [128/200], Step [200/500] Loss: 1.8137
Epoch [128/200], Step [300/500] Loss: 1.3763
Epoch [128/200], Step [400/500] Loss: 1.3240
Epoch [128/200], Step [500/500] Loss: 1.9442
Accuracy of the model on the test images: 75.21 %
use 0.0min20.594711542129517s
Epoch [129/200], Step [100/500] Loss: 1.6730
Epoch [129/200], Step [200/500] Loss: 1.4576
Epoch [129/200], Step [300/500] Loss: 1.2523
Epoch [129/200], Step [400/500] Loss: 1.5209
Epoch [129/200], Step [500/500] Loss: 1.1535
Accuracy of the model on the test images: 75.46 %
use 0.0min20.58636236190796s
Epoch [130/200], Step [100/500] Loss: 1.6338
Epoch [130/200], Step [200/500] Loss: 1.7922
Epoch [130/200], Step [300/500] Loss: 0.5676
Epoch [130/200], Step [400/500] Loss: 1.8949
Epoch [130/200], Step [500/500] Loss: 1.9921
Accuracy of the model on the test images: 75.47 %
use 0.0min20.57331395149231s
Epoch [131/200], Step [100/500] Loss: 0.8292
Epoch [131/200], Step [200/500] Loss: 1.6283
Epoch [131/200], Step [300/500] Loss: 1.0917
Epoch [131/200], Step [400/500] Loss: 1.5508
Epoch [131/200], Step [500/500] Loss: 1.6635
Accuracy of the model on the test images: 75.76 %
use 0.0min20.58394694328308s
Epoch [132/200], Step [100/500] Loss: 0.7177
Epoch [132/200], Step [200/500] Loss: 0.8821
Epoch [132/200], Step [300/500] Loss: 1.2840
Epoch [132/200], Step [400/500] Loss: 1.4966
Epoch [132/200], Step [500/500] Loss: 1.8299
Accuracy of the model on the test images: 75.83 %
use 0.0min20.57688283920288s
Epoch [133/200], Step [100/500] Loss: 1.1117
Epoch [133/200], Step [200/500] Loss: 2.0203
Epoch [133/200], Step [300/500] Loss: 0.2079
Epoch [133/200], Step [400/500] Loss: 1.6799
Epoch [133/200], Step [500/500] Loss: 1.2938
Accuracy of the model on the test images: 75.85 %
use 0.0min20.588330030441284s
Epoch [134/200], Step [100/500] Loss: 1.1838
Epoch [134/200], Step [200/500] Loss: 0.1680
Epoch [134/200], Step [300/500] Loss: 0.6034
Epoch [134/200], Step [400/500] Loss: 1.9377
Epoch [134/200], Step [500/500] Loss: 1.4440
Accuracy of the model on the test images: 75.29 %
use 0.0min20.579824924468994s
Epoch [135/200], Step [100/500] Loss: 1.5189
Epoch [135/200], Step [200/500] Loss: 1.7238
Epoch [135/200], Step [300/500] Loss: 1.8031
Epoch [135/200], Step [400/500] Loss: 1.5680
Epoch [135/200], Step [500/500] Loss: 1.8765
Accuracy of the model on the test images: 75.5 %
use 0.0min20.576337575912476s
Epoch [136/200], Step [100/500] Loss: 1.7640
Epoch [136/200], Step [200/500] Loss: 1.2187
Epoch [136/200], Step [300/500] Loss: 1.6071
Epoch [136/200], Step [400/500] Loss: 1.2763
Epoch [136/200], Step [500/500] Loss: 1.5862
Accuracy of the model on the test images: 75.66 %
use 0.0min20.593910694122314s
Epoch [137/200], Step [100/500] Loss: 1.7075
Epoch [137/200], Step [200/500] Loss: 1.5340
Epoch [137/200], Step [300/500] Loss: 1.4863
Epoch [137/200], Step [400/500] Loss: 1.4170
Epoch [137/200], Step [500/500] Loss: 1.6421
Accuracy of the model on the test images: 75.53 %
use 0.0min20.554625511169434s
Epoch [138/200], Step [100/500] Loss: 1.7390
Epoch [138/200], Step [200/500] Loss: 1.6388
Epoch [138/200], Step [300/500] Loss: 1.3028
Epoch [138/200], Step [400/500] Loss: 1.8569
Epoch [138/200], Step [500/500] Loss: 1.3246
Accuracy of the model on the test images: 75.65 %
use 0.0min20.579295873641968s
Epoch [139/200], Step [100/500] Loss: 0.3075
Epoch [139/200], Step [200/500] Loss: 1.7466
Epoch [139/200], Step [300/500] Loss: 1.3005
Epoch [139/200], Step [400/500] Loss: 0.3846
Epoch [139/200], Step [500/500] Loss: 0.2858
Accuracy of the model on the test images: 75.85 %
use 0.0min20.722463130950928s
Epoch [140/200], Step [100/500] Loss: 0.9956
Epoch [140/200], Step [200/500] Loss: 1.6868
Epoch [140/200], Step [300/500] Loss: 1.2558
Epoch [140/200], Step [400/500] Loss: 1.0439
Epoch [140/200], Step [500/500] Loss: 1.5723
Accuracy of the model on the test images: 75.58 %
use 0.0min20.580142498016357s
Epoch [141/200], Step [100/500] Loss: 2.2597
Epoch [141/200], Step [200/500] Loss: 1.6770
Epoch [141/200], Step [300/500] Loss: 0.8316
Epoch [141/200], Step [400/500] Loss: 1.7355
Epoch [141/200], Step [500/500] Loss: 1.8693
Accuracy of the model on the test images: 75.45 %
use 0.0min20.55280113220215s
Epoch [142/200], Step [100/500] Loss: 1.5666
Epoch [142/200], Step [200/500] Loss: 0.6596
Epoch [142/200], Step [300/500] Loss: 1.6928
Epoch [142/200], Step [400/500] Loss: 1.9183
Epoch [142/200], Step [500/500] Loss: 1.3316
Accuracy of the model on the test images: 75.76 %
use 0.0min20.5780246257782s
Epoch [143/200], Step [100/500] Loss: 1.6893
Epoch [143/200], Step [200/500] Loss: 1.5397
Epoch [143/200], Step [300/500] Loss: 1.6977
Epoch [143/200], Step [400/500] Loss: 1.1024
Epoch [143/200], Step [500/500] Loss: 1.0337
Accuracy of the model on the test images: 75.57 %
use 0.0min20.599243879318237s
Epoch [144/200], Step [100/500] Loss: 0.5643
Epoch [144/200], Step [200/500] Loss: 1.4360
Epoch [144/200], Step [300/500] Loss: 0.7223
Epoch [144/200], Step [400/500] Loss: 0.9717
Epoch [144/200], Step [500/500] Loss: 1.2771
Accuracy of the model on the test images: 75.52 %
use 0.0min20.55806040763855s
Epoch [145/200], Step [100/500] Loss: 1.7184
Epoch [145/200], Step [200/500] Loss: 1.7416
Epoch [145/200], Step [300/500] Loss: 0.9393
Epoch [145/200], Step [400/500] Loss: 1.6941
Epoch [145/200], Step [500/500] Loss: 2.0614
Accuracy of the model on the test images: 75.65 %
use 0.0min20.56918478012085s
Epoch [146/200], Step [100/500] Loss: 1.2707
Epoch [146/200], Step [200/500] Loss: 0.7448
Epoch [146/200], Step [300/500] Loss: 0.9327
Epoch [146/200], Step [400/500] Loss: 1.6580
Epoch [146/200], Step [500/500] Loss: 1.7474
Accuracy of the model on the test images: 75.62 %
use 0.0min20.577457189559937s
Epoch [147/200], Step [100/500] Loss: 0.3349
Epoch [147/200], Step [200/500] Loss: 1.1530
Epoch [147/200], Step [300/500] Loss: 1.7887
Epoch [147/200], Step [400/500] Loss: 1.8637
Epoch [147/200], Step [500/500] Loss: 1.5539
Accuracy of the model on the test images: 75.62 %
use 0.0min20.56734824180603s
Epoch [148/200], Step [100/500] Loss: 1.8549
Epoch [148/200], Step [200/500] Loss: 0.1096
Epoch [148/200], Step [300/500] Loss: 1.6673
Epoch [148/200], Step [400/500] Loss: 1.3853
Epoch [148/200], Step [500/500] Loss: 1.4980
Accuracy of the model on the test images: 75.67 %
use 0.0min20.595983028411865s
Epoch [149/200], Step [100/500] Loss: 0.3207
Epoch [149/200], Step [200/500] Loss: 1.5532
Epoch [149/200], Step [300/500] Loss: 1.4864
Epoch [149/200], Step [400/500] Loss: 1.1584
Epoch [149/200], Step [500/500] Loss: 0.1673
Accuracy of the model on the test images: 75.39 %
use 0.0min20.56972908973694s
Epoch [150/200], Step [100/500] Loss: 1.6175
Epoch [150/200], Step [200/500] Loss: 1.6672
Epoch [150/200], Step [300/500] Loss: 0.2774
Epoch [150/200], Step [400/500] Loss: 0.4501
Epoch [150/200], Step [500/500] Loss: 0.9520
Accuracy of the model on the test images: 75.64 %
use 0.0min20.585551261901855s
Epoch [151/200], Step [100/500] Loss: 1.8260
Epoch [151/200], Step [200/500] Loss: 1.3618
Epoch [151/200], Step [300/500] Loss: 1.5497
Epoch [151/200], Step [400/500] Loss: 1.0226
Epoch [151/200], Step [500/500] Loss: 1.5750
Accuracy of the model on the test images: 75.4 %
use 0.0min20.573359966278076s
Epoch [152/200], Step [100/500] Loss: 1.7537
Epoch [152/200], Step [200/500] Loss: 0.3935
Epoch [152/200], Step [300/500] Loss: 1.3617
Epoch [152/200], Step [400/500] Loss: 0.9150
Epoch [152/200], Step [500/500] Loss: 1.1744
Accuracy of the model on the test images: 75.67 %
use 0.0min20.55489730834961s
Epoch [153/200], Step [100/500] Loss: 1.6436
Epoch [153/200], Step [200/500] Loss: 1.9014
Epoch [153/200], Step [300/500] Loss: 1.7428
Epoch [153/200], Step [400/500] Loss: 1.1601
Epoch [153/200], Step [500/500] Loss: 0.8929
Accuracy of the model on the test images: 75.88 %
use 0.0min20.567218542099s
Epoch [154/200], Step [100/500] Loss: 0.5618
Epoch [154/200], Step [200/500] Loss: 1.4692
Epoch [154/200], Step [300/500] Loss: 1.1863
Epoch [154/200], Step [400/500] Loss: 1.2290
Epoch [154/200], Step [500/500] Loss: 0.1978
Accuracy of the model on the test images: 76.0 %

你可能感兴趣的:(笔记,深度学习,神经网络,pytorch)

  • 机器学习与深度学习间关系与区别 ℒℴѵℯ心·动ꦿ໊ོ꫞ 人工智能学习深度学习python
    一、机器学习概述定义机器学习（MachineLearning,ML）是一种通过数据驱动的方法，利用统计学和计算算法来训练模型，使计算机能够从数据中学习并自动进行预测或决策。机器学习通过分析大量数据样本，识别其中的模式和规律，从而对新的数据进行判断。其核心在于通过训练过程，让模型不断优化和提升其预测准确性。主要类型1.监督学习（SupervisedLearning）监督学习是指在训练数据集中包含输入
  • 10月|愿你的青春不负梦想-读书笔记-01 Tracy的小书斋
    本书的作者是俞敏洪，大家都很熟悉他了吧。俞敏洪老师是我行业的领头羊吧，也是我事业上的偶像。本日摘录他书中第一章中的金句：『一个人如果什么目标都没有，就会浑浑噩噩，感觉生命中缺少能量。能给我们能量的，是对未来的期待。第一件事，我始终为了进步而努力。与其追寻全世界的骏马，不如种植丰美的草原，到时骏马自然会来。第二件事，我始终有阶段性的目标。什么东西能给我能量？答案是对未来的期待。』读到这里的时候，我便
  • 《投行人生》读书笔记 小蘑菇的树洞
    《投行人生》----作者詹姆斯-A-朗德摩根斯坦利副主席40年的职业洞见-很短小精悍的篇幅，比较适合初入职场的新人。第一部分成功的职业生涯需要规划1.情商归为适应能力分享与协作同理心适应能力，更多的是自我意识，你有能力识别自己的情并分辨这些情绪如何影响你的思想和行为。2.对于初入职场的人的建议，细节，截止日期和数据很重要截止日期，一种有效的方法是请老板为你所有的任务进行优先级排序。和老板喝咖啡的好
  • 【一起学Rust | 设计模式】习惯语法——使用借用类型作为参数、格式化拼接字符串、构造函数 广龙宇 一起学Rust#Rust设计模式rust设计模式开发语言
    提示：文章写完后，目录可以自动生成，如何生成可参考右边的帮助文档文章目录前言一、使用借用类型作为参数二、格式化拼接字符串三、使用构造函数总结前言Rust不是传统的面向对象编程语言，它的所有特性，使其独一无二。因此，学习特定于Rust的设计模式是必要的。本系列文章为作者学习《Rust设计模式》的学习笔记以及自己的见解。因此，本系列文章的结构也与此书的结构相同（后续可能会调成结构），基本上分为三个部分
  • git常用命令笔记 咩酱-小羊 git笔记
    ###用习惯了idea总是不记得git的一些常见命令，需要用到的时候总是担心旁边站了人~~~记个笔记@_@，告诉自己看笔记不丢人初始化初始化一个新的Git仓库gitinit配置配置用户信息gitconfig--globaluser.name"YourName"gitconfig--globaluser.email"[email protected]"基本操作克隆远程仓库gitclone查看
  • 将cmd中命令输出保存为txt文本文件 落难Coder Windowscmdwindow
    最近深度学习本地的训练中我们常常要在命令行中运行自己的代码，无可厚非，我们有必要保存我们的炼丹结果，但是复制命令行输出到txt是非常麻烦的，其实Windows下的命令行为我们提供了相应的操作。其基本的调用格式就是：运行指令>输出到的文件名称或者具体保存路径测试下，我打开cmd并且ping一下百度：pingwww.baidu.com>./data.txt看下相同目录下data.txt的输出：如果你再
  • 509. 斐波那契数(每日一题) lzyprime
    lzyprime博客(github)创建时间：2021.01.04qq及邮箱：2383518170leetcode笔记题目描述斐波那契数，通常用F(n)表示，形成的序列称为斐波那契数列。该数列由0和1开始，后面的每一项数字都是前面两项数字的和。也就是：F(0)=0，F(1)=1F(n)=F(n-1)+F(n-2)，其中n>1给你n，请计算F(n)。示例1：输入：2输出：1解释：F(2)=F(1)+
  • 拥有断舍离的心态，过精简生活--《断舍离》读书笔记 爱吃丸子的小樱桃
    不知不觉间房间里的东西越来越多，虽然摆放整齐，但也时常会觉得空间逼仄，令人心生烦闷。抱着断舍离的态度，我开始阅读《断舍离》这本书，希望从书中能找到一些有效的方法，帮助我实现空间、物品上的断舍离。《断舍离》是日本作家山下英子通过自己的经历、思考和实践总结而成的，整体内涵也从刚开始的私人生活哲学的“断舍离”升华成了“人生实践哲学”，接着又成为每个人都能实行的“改变人生的断舍离”，从“哲学”逐渐升华成“
  • 四章-32-点要素的聚合 彩云飘过
    本文基于腾讯课堂老胡的课《跟我学Openlayers--基础实例详解》做的学习笔记，使用的openlayers5.3.xapi。源码见1032.html，对应的官网示例https://openlayers.org/en/latest/examples/cluster.htmlhttps://openlayers.org/en/latest/examples/earthquake-clusters.
  • 高端密码学院笔记285 柚子_b4b4
    高端幸福密码学院（高级班）幸福使者：李华第（598）期《幸福》之回归内在深层生命原动力基础篇——揭秘“激励”成长的喜悦心理案例分析主讲：刘莉一，知识扩充:成功=艰苦劳动+正确方法+少说空话。贪图省力的船夫，目标永远下游。智者的梦再美，也不如愚人实干的脚印。幸福早课堂2020.10.16星期五一笔记:1，重视和珍惜的前提是知道它的价值非常重要，当你珍惜了，你就真正定下来，真正的学到身上。2，大家需要
  • Day17笔记-高阶函数 ~在杰难逃~ Python笔记python开发语言pycharm数据分析
    高阶函数【重点掌握】函数的本质：函数是一个变量，函数名是一个变量名，一个函数可以作为另一个函数的参数或返回值使用如果A函数作为B函数的参数，B函数调用完成之后，会得到一个结果，则B函数被称为高阶函数常用的高阶函数：map(),reduce(),filter(),sorted()1.map()map(func,iterable)，返回值是一个iterator【容器，迭代器】func:函数iterab
  • Day1笔记-Python简介&标识符和关键字&输入输出 ~在杰难逃~ Pythonpython开发语言大数据数据分析数据挖掘
    大家好，从今天开始呢，杰哥开展一个新的专栏，当然，数据分析部分也会不定时更新的，这个新的专栏主要是讲解一些Python的基础语法和知识，帮助0基础的小伙伴入门和学习Python，感兴趣的小伙伴可以开始认真学习啦！一、Python简介【了解】1.计算机工作原理编程语言就是用来定义计算机程序的形式语言。我们通过编程语言来编写程序代码，再通过语言处理程序执行向计算机发送指令，让计算机完成对应的工作，编程
  • node.js学习 小猿L node.jsnode.js学习vim
    node.js学习实操及笔记温故node.js，node.js学习实操过程及笔记~node.js学习视频node.js官网node.js中文网实操笔记githubcsdn笔记为什么学node.js可以让别人访问我们编写的网页为后续的框架学习打下基础，三大框架vuereactangular离不开node.jsnode.js是什么官网：node.js是一个开源的、跨平台的运行JavaScript的运行
  • 数据仓库——维度表一致性 墨染丶eye 背诵数据仓库
    数据仓库基础笔记思维导图已经整理完毕，完整连接为：数据仓库基础知识笔记思维导图维度一致性问题从逻辑层面来看，当一系列星型模型共享一组公共维度时，所涉及的维度称为一致性维度。当维度表存在不一致时，短期的成功难以弥补长期的错误。维度时确保不同过程中信息集成起来实现横向钻取货活动的关键。造成横向钻取失败的原因维度结构的差别，因为维度的差别，分析工作涉及的领域从简单到复杂，但是都是通过复杂的报表来弥补设计
  • 【Git】常见命令(仅笔记) 好想有猫猫 GitLinux学习笔记git笔记elasticsearchlinuxc++
    文章目录创建/初始化本地仓库添加本地仓库配置项提交文件查看仓库状态回退仓库查看日志分支删除文件暂存工作区代码远程仓库使用`.gitigore`文件让git不追踪一些文件标签创建/初始化本地仓库gitinit添加本地仓库配置项gitconfig-l#以列表形式显示配置项gitconfiguser.name"ljh"#配置user.namegitconfiguser.email"[email protected]
  • 为什么你总是对下属不满意? ZhaoWu1050
    【ZhaoWu的听课笔记】大多数公司，都存在两种问题。我创业四年，更是体会深切。这两种问题就是：老板经常不满意下属的表现；下属总是不知道老板想要什么；虽然这两种问题普遍存在，其实解决方法并不复杂。这节课，我们再聊聊第一个问题：为什么老板经常不满意下属表现?其实，这背后也是一条管理常识。管理学家德鲁克先生早就说过：管理者的任务，不是去改变人。*来自《卓有成效的管理者》只是大多数老板和我一样，都是一边
  • 母亲节如何做小红书营销 美橙传媒
    小红书的一举一动引起了外界的高度关注。通过爆款笔记和流行话题，我们可以看到“干货”类型的内容在小红书中偏向实用的生活经验共享和生活指南非常受欢迎。根据运营社的分析，这种现象是由小红书用户心智和内容社区背后机制共同决定的。首先，小红书将使用“强搜索”逻辑为用户提供特定的“搜索场景”。在“我必须这样生活”中，大量使用了满足小红书站用户喜好和需求的内容。内容社区自制的高质量内容也吸引了寻找营销新途径的品
  • 读书笔记|《遇见孩子，遇见更好的自己》5 抹茶社长
    为人父母意味着放弃自己的过去，不要对以往没有实现的心愿耿耿于怀，只有这样，孩子们才能做回自己。985909803.jpg孩子在与父母保持亲密的同时更需要独立，唯有这样，孩子才会成为孩子，父母才会成其为父母。有耐心的人生往往更幸福，给孩子留点余地。认识到养儿育女是对耐心的考验。为失败做好心理准备，教会孩子控制情绪。了解自己的底线，说到底线，有一点很重要，父母之所以发脾气，真正的原因往往在于他们自己，
  • 基于Python给出的PDF文档转Markdown文档的方法 程序媛了了 pythonpdf开发语言
    注：网上有很多将Markdown文档转为PDF文档的方法，但是却很少有将PDF文档转为Markdown文档的方法。就算有，比如某些网站声称可以将PDF文档转为Markdown文档，尝试过，不太符合自己的要求，而且无法保证文档没有泄露风险。于是本人为了解决这个问题，借助GPT（能使用GPT镜像或者有条件直接使用GPT的，反正能调用GPT接口就行）生成Python代码来完成这个功能。笔记、代码难免存在
  • 语文主题教学学习笔记之87 东哥杂谈
    “语文主题教学”学习笔记之八十七（0125）今天继续学习小学语文主题教学的实践样态。板块三：教学中体现“书艺”味道。作为四大名著之一的《水浒传》，堪称我国文学宝库之经典。对从《水浒传》中摘选的单元，教师就要了解其原生态，即评书体特点。这也要求教师要了解一些常用的评书行话术语，然后在教学时适时地加入一些，让学生体味其文本中原有的特色。学生也要尽可能地通过朗读的方式，而不单是分析讲解的方式进行学习。细
  • Armv8.3 体系结构扩展--原文版 代码改变世界ctw ARM-TEE-Androidarmv8嵌入式arm架构安全架构芯片TrustzoneSecureboot
    快速链接:.ARMv8/ARMv9架构入门到精通-[目录]付费专栏-付费课程【购买须知】:个人博客笔记导读目录(全部)TheArmv8.3architectureextensionTheArmv8.3architectureextensionisanextensiontoArmv8.2.Itaddsmandatoryandoptionalarchitecturalfeatures.Somefeat
  • springboot+vue项目实战一-创建SpringBoot简单项目 苹果酱0567 面试题汇总与解析springboot后端java中间件开发语言
    这段时间抽空给女朋友搭建一个个人博客，想着记录一下建站的过程，就当做笔记吧。虽然复制zjblog只要一个小时就可以搞定一个网站，或者用cms系统，三四个小时就可以做出一个前后台都有的网站，而且想做成啥样也都行。但是就是要从新做，自己做的意义不一样，更何况，俺就是专门干这个的，嘿嘿嘿要做一个网站，而且从零开始，首先呢就是技术选型了，经过一番思量决定选择-SpringBoot做后端，前端使用Vue做一
  • 阅读《认知觉醒》读书笔记 就看看书
    本周阅读了周岭的《认知觉醒开启自我改变的原动力》，启发较多，故做读书笔记一则，留待学习。全书共八章，讲述了大脑、潜意识、元认知、专注力、学习力、行动力、情绪力及成本最低的成长之道。具体描述了大脑、焦虑、耐心、模糊、感性、元认知、自控力、专注力、情绪专注、学习专注、匹配、深度、关联、体系、打卡、反馈、休息、清晰、傻瓜、行动、心智宽带、单一视角、游戏心态、早起、冥想、阅读、写作、运动等相关知识点。大脑
  • 阅读笔记：阅读方法中的逻辑和转念 施吉涛
    聊聊一些阅读的方法论吧，别人家的读书方法刚开始想写，然后就不知道写什么了，因为作者写的非常的“精致”我有一种乡巴佬进城的感觉，看到精美的摆盘，精致的食材不知道该如何下口也就是《阅读的方法》，我们姑且来试一下强劲的大脑篇，第一节：逻辑通俗的来讲，也就是表达的排列和顺序，再进一步就是因果关系和关联实际上书已经看了大概一遍，但直到打算写一下笔记的时候，才发现作者讲的推理更多的是阅读的对象中呈现出的逻辑也
  • 《转介绍方法论》学习笔记 小可乐的妈妈
    一、高效转介绍的流程：价值观---执行----方案一）转介绍发生的背景：1、对象：谁向谁转介绍？全员营销，人人参与。①员工的激励政策、客户的转介绍诱因制作客户画像：a信任；支付能力；意愿度；便利度（根据家长具备四个特征的个数分为四类）B性格分类C职业分类D年龄性别②执行：套路，策略，方法，流程2、诱因：为什么要转介绍？认同信任；多方共赢；传递美好；零风险承诺打动人心，超越期待。选择做教育，就是选择
  • JAVA学习笔记之23种设计模式学习 victorfreedom Java技术设计模式androidjava常用设计模式
    博主最近买了《设计模式》这本书来学习，无奈这本书是以C++语言为基础进行说明，整个学习流程下来效率不是很高，虽然有的设计模式通俗易懂，但感觉还是没有充分的掌握了所有的设计模式。于是博主百度了一番，发现有大神写过了这方面的问题，于是博主迅速拿来学习。一、设计模式的分类总体来说设计模式分为三大类：创建型模式，共五种：工厂方法模式、抽象工厂模式、单例模式、建造者模式、原型模式。结构型模式，共七种：适配器
  • 解决Obsidian写笔记中的＜img＞标签无法显示图片的问题 全能全知者 笔记
    Obsidian中写md笔记如果使用标签会显示不出图案，后来才知道因为Obsidian的问题导致只能用绝对路径定位。所以我本人写了一个py插件，将md笔记里的img标签批量替换成Obsidian能够读取的形式。安装FixObsImgDpy:pipinstallFixObsImgDpy安装完成后在需要修复的md文件的父目录下运行命令:FixObsImgDpy就会自动修复父目录以下的全部md文件 仓库
  • 2021年周总结 03 Ruby之家
    这周的生活过得也是比较快，因为暂时住的离公司有点距离，所以通勤时间相对较长一点，而在地铁上的一个半小时如何充分利用起来，则是我最近一直在思考的问题，2021年想让自己的生活都运行在计划中。(有时候自己想干一件事情就总是给自己找很多借口，想着以后怎么怎么样？然而哪有那么多的以后，能够方便当下的工作生活就立马执行就OK，这仅仅只是我此时想到背的很重的老人机笔记本电脑，也算是陪伴我快8年的—当时买的时候
  • 推荐3家毕业AI论文可五分钟一键生成！文末附免费教程！ 小猪包333 写论文人工智能AI写作深度学习计算机视觉
    在当前的学术研究和写作领域，AI论文生成器已经成为许多研究人员和学生的重要工具。这些工具不仅能够帮助用户快速生成高质量的论文内容，还能进行内容优化、查重和排版等操作。以下是三款值得推荐的AI论文生成器：千笔-AIPassPaper、懒人论文以及AIPaperPass。千笔-AIPassPaper千笔-AIPassPaper是一款基于深度学习和自然语言处理技术的AI写作助手，旨在帮助用户快速生成高质
  • AI大模型的架构演进与最新发展 季风泯灭的季节 AI大模型应用技术二人工智能架构
    随着深度学习的发展，AI大模型（LargeLanguageModels,LLMs）在自然语言处理、计算机视觉等领域取得了革命性的进展。本文将详细探讨AI大模型的架构演进，包括从Transformer的提出到GPT、BERT、T5等模型的历史演变，并探讨这些模型的技术细节及其在现代人工智能中的核心作用。一、基础模型介绍：Transformer的核心原理Transformer架构的背景在Transfo
  • log4j对象改变日志级别 3213213333332132 javalog4jlevellog4j对象名称日志级别
    log4j对象改变日志级别可批量的改变所有级别，或是根据条件改变日志级别。 log4j配置文件： log4j.rootLogger=ERROR,FILE,CONSOLE,EXECPTION #log4j.appender.FILE=org.apache.log4j.RollingFileAppender log4j.appender.FILE=org.apache.l
  • elk+redis 搭建nginx日志分析平台 ronin47 elasticsearchkibanalogstash
                 elk+redis 搭建nginx日志分析平台 logstash,elasticsearch,kibana 怎么进行nginx的日志分析呢？首先，架构方面，nginx是有日志文件的，它的每个请求的状态等都有日志文件进行记录。其次，需要有个队 列，redis的l
  • Yii2设置时区 dcj3sjt126com PHPtimezoneyii2
    时区这东西，在开发的时候，你说重要吧，也还好，毕竟没它也能正常运行，你说不重要吧，那就纠结了。特别是linux系统，都TMD差上几小时，你能不痛苦吗？win还好一点。有一些常规方法，是大家目前都在采用的1、php.ini中的设置，这个就不谈了，2、程序中公用文件里设置，date_default_timezone_set一下时区3、或者。。。自己写时间处理函数，在遇到时间的时候，用这个函数处理（比较
  • js实现前台动态添加文本框，后台获取文本框内容 171815164 文本框
    <%@ page language="java" import="java.util.*" pageEncoding="UTF-8"%> <!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN" "http://w
  • 持续集成工具 g21121 持续集成
            持续集成是什么？我们为什么需要持续集成？持续集成带来的好处是什么？什么样的项目需要持续集成？...        持续集成(Continuous integration ,简称CI)，所谓集成可以理解为将互相依赖的工程或模块合并成一个能单独运行
  • 数据结构哈希表(hash)总结 永夜-极光 数据结构
    1.什么是hash 来源于百度百科: Hash，一般翻译做“散列”，也有直接音译为“哈希”的，就是把任意长度的输入，通过散列算法，变换成固定长度的输出，该输出就是散列值。这种转换是一种压缩映射，也就是，散列值的空间通常远小于输入的空间，不同的输入可能会散列成相同的输出，所以不可能从散列值来唯一的确定输入值。简单的说就是一种将任意长度的消息压缩到某一固定长度的消息摘要的函数。  
  • 乱七八糟 程序员是怎么炼成的
    eclipse中的jvm字节码查看插件地址： http://andrei.gmxhome.de/eclipse/ 安装该地址的outline 插件  后重启，打开window下的view下的bytecode视图 http://andrei.gmxhome.de/eclipse/   jvm博客： http://yunshen0909.iteye.com/blog/2
  • 职场人伤害了“上司” 怎样弥补 aijuans 职场
    由于工作中的失误，或者平时不注意自己的言行“伤害”、“得罪”了自己的上司，怎么办呢？ 　　在职业生涯中这种问题尽量不要发生。下面提供了一些解决问题的建议： 　　一、利用一些轻松的场合表示对他的尊重 　　即使是开明的上司也很注重自己的权威，都希望得到下属的尊重，所以当你与上司冲突后，最好让不愉快成为过去，你不妨在一些轻松的场合，比如会餐、联谊活动等，向上司问个好，敬下酒，表示你对对方的尊重，
  • 深入浅出url编码 antonyup_2006 应用服务器浏览器servletweblogicIE
    出处：http://blog.csdn.net/yzhz  杨争   http://blog.csdn.net/yzhz/archive/2007/07/03/1676796.aspx 一、问题：         编码问题是JAVA初学者在web开发过程中经常会遇到问题，网上也有大量相关的
  • 建表后创建表的约束关系和增加表的字段 百合不是茶 标的约束关系增加表的字段
      下面所有的操作都是在表建立后操作的,主要目的就是熟悉sql的约束,约束语句的万能公式   1,增加字段(student表中增加 姓名字段)   alter table 增加字段的表名 add 增加的字段名 增加字段的数据类型 alter table student add name varchar2(10);   &nb
  • Uploadify 3.2 参数属性、事件、方法函数详解 bijian1013 JavaScriptuploadify
    一.属性 属性名称 默认值 说明 auto true 设置为true当选择文件后就直接上传了，为false需要点击上传按钮才上传。 buttonClass ” 按钮样式 buttonCursor ‘hand’ 鼠标指针悬停在按钮上的样子 buttonImage null 浏览按钮的图片的路
  • 精通Oracle10编程SQL(16)使用LOB对象 bijian1013 oracle数据库plsql
    /* *使用LOB对象 */ --LOB(Large Object)是专门用于处理大对象的一种数据类型，其所存放的数据长度可以达到4G字节 --CLOB/NCLOB用于存储大批量字符数据，BLOB用于存储大批量二进制数据，而BFILE则存储着指向OS文件的指针 /* *综合实例 */ --建立表空间 --#指定区尺寸为128k,如不指定，区尺寸默认为64k CR
  • 【Resin一】Resin服务器部署web应用 bit1129 resin
    工作中，在Resin服务器上部署web应用，通常有如下三种方式：   配置多个web-app 配置多个http id 为每个应用配置一个propeties、xml以及sh脚本文件 配置多个web-app   在resin.xml中,可以为一个host配置多个web-app   <cluster id="app&q
  • red5简介及基础知识 白糖_ 基础
      简介   Red5的主要功能和Macromedia公司的FMS类似，提供基于Flash的流媒体服务的一款基于Java的开源流媒体服务器。它由Java语言编写，使用RTMP作为流媒体传输协议，这与FMS完全兼容。它具有流化FLV、MP3文件，实时录制客户端流为FLV文件，共享对象，实时视频播放、Remoting等功能。用Red5替换FMS后,客户端不用更改可正
  • angular.fromJson boyitech AngularJSAngularJS 官方APIAngularJS API
    angular.fromJson 描述: 把Json字符串转为对象 使用方法: angular.fromJson(json); 参数详解: Param Type Details json string JSON 字符串 返回值: 对象, 数组, 字符串 或者是一个数字 示例: <!DOCTYPE HTML> <h
  • java-颠倒一个句子中的词的顺序。比如： I am a student颠倒后变成：student a am I bylijinnan java
    public class ReverseWords { /** * 题目：颠倒一个句子中的词的顺序。比如： I am a student颠倒后变成：student a am I.词以空格分隔。 * 要求： * 1.实现速度最快,移动最少 * 2.不能使用String的方法如split,indexOf等等。 * 解答：两次翻转。 */ publ
  • web实时通讯 Chen.H Web浏览器socket脚本
    关于web实时通讯，做一些监控软件。 由web服务器组件从消息服务器订阅实时数据，并建立消息服务器到所述web服务器之间的连接，web浏览器利用从所述web服务器下载到web页面的客户端代理与web服务器组件之间的socket连接，建立web浏览器与web服务器之间的持久连接；利用所述客户端代理与web浏览器页面之间的信息交互实现页面本地更新，建立一条从消息服务器到web浏览器页面之间的消息通路
  • [基因与生物]远古生物的基因可以嫁接到现代生物基因组中吗? comsci 生物
          大家仅仅把我说的事情当作一个IT行业的笑话来听吧..没有其它更多的意思     如果我们把大自然看成是一位伟大的程序员,专门为地球上的生态系统编制基因代码,并创造出各种不同的生物来,那么6500万年前的程序员开发的代码,是否兼容现代派的程序员的代码和架构呢?   
  • oracle 外部表 daizj oracle外部表external tables
        oracle外部表是只允许只读访问，不能进行DML操作，不能创建索引，可以对外部表进行的查询，连接，排序，创建视图和创建同义词操作。 you can select, join, or sort external table data. You can also create views and synonyms for external tables. Ho
  • aop相关的概念及配置 daysinsun AOP
    切面(Aspect): 通常在目标方法执行前后需要执行的方法（如事务、日志、权限），这些方法我们封装到一个类里面，这个类就叫切面。 连接点（joinpoint） spring里面的连接点指需要切入的方法，通常这个joinpoint可以作为一个参数传入到切面的方法里面（非常有用的一个东西）。 通知（Advice） 通知就是切面里面方法的具体实现，分为前置、后置、最终、异常环
  • 初一上学期难记忆单词背诵第二课 dcj3sjt126com englishword
    middle 中间的，中级的 well 喔，那么；好吧 phone 电话，电话机 policeman 警察 ask 问 take 拿到；带到 address 地址 glad 高兴的，乐意的 why 为什么  China 中国 family 家庭 grandmother (外)祖母 grandfather (外)祖父 wife 妻子 husband 丈夫 da
  • Linux日志分析常用命令 dcj3sjt126com linuxlog
    1.查看文件内容 cat -n 显示行号 2.分页显示 more Enter 显示下一行 空格 显示下一页 F 显示下一屏 B 显示上一屏 less /get 查询"get"字符串并高亮显示 3.显示文件尾 tail -f 不退出持续显示 -n 显示文件最后n行 4.显示头文件 head -n 显示文件开始n行 5.内容排序 sort -n 按照
  • JSONP 原理分析 fantasy2005 JavaScriptjsonpjsonp 跨域
    转自 http://www.nowamagic.net/librarys/veda/detail/224 JavaScript是一种在Web开发中经常使用的前端动态脚本技术。在JavaScript中，有一个很重要的安全性限制，被称为“Same-Origin Policy”（同源策略）。这一策略对于JavaScript代码能够访问的页面内容做了很重要的限制，即JavaScript只能访问与包含它的
  • 使用connect by进行级联查询 234390216 oracle查询父子Connect by级联
    使用connect by进行级联查询          connect by可以用于级联查询，常用于对具有树状结构的记录查询某一节点的所有子孙节点或所有祖辈节点。          来看一个示例，现假设我们拥有一个菜单表t_menu，其中只有三个字段：
  • 一个不错的能将HTML表格导出为excel,pdf等的jquery插件 jackyrong jquery插件
    发现一个老外写的不错的jquery插件，可以实现将HTML 表格导出为excel,pdf等格式， 地址在： https://github.com/kayalshri/ 下面看个例子，实现导出表格到excel,pdf <html> <head> <title>Export html table to excel an
  • UI设计中我们为什么需要设计动效 lampcy UIUI设计
    关于Unity3D中的Shader的知识 首先先解释下Unity3D的Shader，Unity里面的Shaders是使用一种叫ShaderLab的语言编写的，它同微软的FX文件或者NVIDIA的CgFX有些类似。传统意义上的vertex shader和pixel shader还是使用标准的Cg/HLSL 编程语言编写的。因此Unity文档里面的Shader，都是指用ShaderLab编写的代码，
  • 如何禁止页面缓存 nannan408 htmljspcache
    禁止页面使用缓存~ ------------------------------------------------ jsp:页面no cache： response.setHeader("Pragma","No-cache"); response.setHeader("Cache-Control","no-cach
  • 以代码的方式管理quartz定时任务的暂停、重启、删除、添加等 Everyday都不同 定时任务管理spring-quartz
          【前言】在项目的管理功能中，对定时任务的管理有时会很常见。因为我们不能指望只在配置文件中配置好定时任务就行了，因为如果要控制定时任务的 “暂停” 呢？暂停之后又要在某个时间点 “重启” 该定时任务呢？或者说直接 “删除” 该定时任务呢？要改变某定时任务的触发时间呢？ “添加” 一个定时任务对于系统的使用者而言，是不太现实的，因为一个定时任务的处理逻辑他是不
  • EXT实例 tntxia ext
      （1） 增加一个按钮   JSP:   <%@ page language="java" import="java.util.*" pageEncoding="UTF-8"%> <% String path = request.getContextPath(); Stri
  • 数学学习在计算机研究领域的作用和重要性 xjnine Math
    最近一直有师弟师妹和朋友问我数学和研究的关系，研一要去学什么数学课。毕竟在清华，衡量一个研究生最重要的指标之一就是paper,而没有数学，是肯定上不了世界顶级的期刊和会议的，这在计算机学界尤其重要！你会发现，不论哪个领域有价值的东西，都一定离不开数学！在这样一个信息时代，当google已经让世界没有秘密的时候，一种卓越的数学思维，绝对可以成为你的核心竞争力.  无奈本人实在见地
按字母分类： ABCDEFGHIJKLMNOPQRSTUVWXYZ其他