10-shuffleNet 图像分类
shuffleNet的pytoch实现：
1 import torch.nn as nn 2 import torch 3 import torch.nn.functional as F 4 from collections import OrderedDict 5 # from torchsummary import summary 6 7 8 def _make_divisible(ch, divisor=8, min_ch=None): 9 """ 10 This function is taken from the original tf repo. 11 It ensures that all layers have a channel number that is divisible by 8 12 It can be seen here: 13 https://github.com/tensorflow/models/blob/master/research/slim/nets/mobilenet/mobilenet.py 14 """ 15 if min_ch is None: 16 min_ch = divisor 17 new_ch = max(min_ch, int(ch + divisor / 2) // divisor * divisor) 18 # Make sure that round down does not go down by more than 10%. 19 if new_ch < 0.9 * ch: 20 new_ch += divisor 21 return new_ch 22 23 #基本的卷积BNrelu 24 class baseConv(nn.Module): 25 def __init__(self,inchannels,outchannels,kernel_size,stride,groups,hasRelu=False): 26 super(baseConv, self).__init__() 27 if hasRelu: 28 #判断是否有relu激活函数 29 activate=nn.ReLU 30 else: 31 activate=nn.Identity 32 pad=kernel_size//2 33 self.baseconv=nn.Sequential( 34 nn.Conv2d(in_channels=inchannels,out_channels=outchannels,kernel_size=kernel_size,stride=stride,padding=pad,groups=groups,bias=False), 35 nn.BatchNorm2d(outchannels), 36 activate() 37 ) 38 39 def forward(self,x): 40 out=self.baseconv(x) 41 return out 42 43 44 #通道重排 45 def ChannelShuffle(x,groups): 46 batch_size,channel,height,width=x.size() 47 #获得每组的组内channel 48 inner_channel=channel//groups 49 #[batch,groups,inner_channel,height,width] 50 x=x.view(batch_size,groups,inner_channel,height,width) 51 x=torch.transpose(x,1,2).contiguous() 52 x=x.view(batch_size,-1,height,width) 53 return x 54 55 56 #stage结构 57 class Residual(nn.Module): 58 def __init__(self,inchannels,outchannels,stride,groups): 59 super(Residual, self).__init__() 60 self.add_=True #shortcut为相加操作 61 self.groups=groups 62 63 hidden_channel=inchannels//4 64 #当输入channel不等于24时候才有第一个1*1conv 65 self.has_conv1=True 66 if inchannels!=24: 67 self.channel1_first1=baseConv(inchannels=inchannels,outchannels=hidden_channel,kernel_size=1,stride=1,groups=groups,hasRelu=True) 68 else: 69 self.has_conv1=False 70 self.channel1_first1=nn.Identity() 71 hidden_channel=inchannels 72 73 #channel1 74 self.channel1=nn.Sequential( 75 baseConv(inchannels=hidden_channel,outchannels=hidden_channel,kernel_size=3,stride=stride,groups=hidden_channel), 76 baseConv(inchannels=hidden_channel,outchannels=outchannels,kernel_size=1,stride=1,groups=groups) 77 ) 78 79 #channel2 80 if stride==2: 81 self.channel2=nn.AvgPool2d(kernel_size=3,stride=stride,padding=1) 82 self.add_=False 83 84 85 def forward(self,x): 86 if self.has_conv1: 87 x1=self.channel1_first1(x) 88 x1=ChannelShuffle(x1,groups=self.groups) 89 out=self.channel1(x1) 90 else: 91 out=self.channel1(x) 92 if self.add_: 93 out+=x 94 return F.relu_(out) 95 else: 96 out2=self.channel2(x) 97 out=torch.cat((out,out2),dim=1) 98 return F.relu_(out) 99 100 101 #shuffleNet 102 class ShuffleNet(nn.Module): 103 def __init__(self,groups,out_channel_list,num_classes,rate,init_weight=True): 104 super(ShuffleNet, self).__init__() 105 106 #定义有序字典存放网络结构 107 self.Module_List=OrderedDict() 108 109 self.Module_List.update({'Conv1':nn.Sequential(nn.Conv2d(3,_make_divisible(24*rate,divisor=4*groups),3,2,1,bias=False),nn.BatchNorm2d(_make_divisible(24*rate,4*groups)),nn.ReLU())}) 110 self.Module_List.update({'MaxPool1':nn.MaxPool2d(3,2,1)}) 111 112 #net_config [inchannels,outchannels,stride] 113 net_config=[[out_channel_list[0],out_channel_list[0],1], 114 [out_channel_list[0],out_channel_list[1],2], 115 [out_channel_list[1],out_channel_list[2],1], 116 [out_channel_list[2],out_channel_list[3],2], 117 [out_channel_list[3],out_channel_list[4],1]] 118 repeat_num=[3,1,7,1,3] 119 120 #搭建stage部分 121 self.Module_List.update({'stage0_0':Residual(_make_divisible(24*rate,4*groups),_make_divisible((out_channel_list[0]-_make_divisible(24*rate,4*groups))*rate,4*groups),stride=2,groups=groups)}) 122 for idx,item in enumerate(repeat_num): 123 config_item=net_config[idx] 124 for j in range(item): 125 if j==0 and idx!=0 and config_item[-1]==2: 126 self.Module_List.update({'stage{}_{}'.format(idx,j+1):Residual(_make_divisible(config_item[0]*rate,4*groups),_make_divisible((config_item[1]-config_item[0])*rate,4*groups),config_item[2],groups)}) 127 else: 128 self.Module_List.update({'stage{}_{}'.format(idx,j+1):Residual(_make_divisible(config_item[0]*rate,4*groups),_make_divisible(config_item[1]*rate,4*groups),config_item[2],groups)}) 129 config_item[-1]=1 #重复stage的stride=1 130 config_item[0]=config_item[1] 131 132 self.Module_List.update({'GlobalPool':nn.AvgPool2d(kernel_size=7,stride=1)}) 133 134 self.Module_List=nn.Sequential(self.Module_List) 135 136 self.linear=nn.Sequential( 137 nn.Dropout(p=0.2), 138 nn.Linear(_make_divisible(out_channel_list[-1]*rate,4*groups),num_classes) 139 ) 140 141 if init_weight: 142 self.init_weight() 143 def forward(self,x): 144 out=self.Module_List(x) 145 out=out.view(out.size(0),-1) 146 out=self.linear(out) 147 return out 148 149 def init_weight(self): 150 for w in self.modules(): 151 if isinstance(w, nn.Conv2d): 152 nn.init.kaiming_normal_(w.weight, mode='fan_out') 153 if w.bias is not None: 154 nn.init.zeros_(w.bias) 155 elif isinstance(w, nn.BatchNorm2d): 156 nn.init.ones_(w.weight) 157 nn.init.zeros_(w.bias) 158 elif isinstance(w, nn.Linear): 159 nn.init.normal_(w.weight, 0, 0.01) 160 nn.init.zeros_(w.bias) 161 162 #定义shufflenet_ 163 def shuffleNet_g1_(num_classes=10,rate=1.0): 164 config=[144,288,288,576,576] 165 return ShuffleNet(groups=1,out_channel_list=config,num_classes=num_classes,rate=rate) 166 167 def shuffleNet_g2_(num_classes=10,rate=1.0): # 168 config=[200,400,400,800,800] 169 return ShuffleNet(groups=2,out_channel_list=config,num_classes=num_classes,rate=rate) 170 171 def shuffleNet_g3_(num_classes=10,rate=1.0): 172 config=[240,480,480,960,960] 173 return ShuffleNet(groups=3,out_channel_list=config,num_classes=num_classes,rate=rate) 174 175 def shuffleNet_g4_(num_classes=10,rate=1.0): 176 config=[272,544,544,1088,1088] 177 return ShuffleNet(groups=4,out_channel_list=config,num_classes=num_classes,rate=rate) 178 179 def shuffleNet_g8_(num_classes=10,rate=1.0): 180 config=[384,768,768,1536,1536] 181 return ShuffleNet(groups=8,out_channel_list=config,num_classes=num_classes,rate=rate) 182 183 # if __name__ == '__main__': 184 # net=shuffleNet_g3_(10,rate=1.0).to('cuda') 185 # print(net) 186 # summary(net,(3,224,224))
classfyNet_train.py
1 import torch 2 from torch.utils.data import DataLoader 3 from torch import nn, optim 4 from torchvision import datasets, transforms 5 from torchvision.transforms.functional import InterpolationMode 6 7 from matplotlib import pyplot as plt 8 9 10 import time 11 12 from Lenet5 import Lenet5_new 13 from Resnet18 import ResNet18,ResNet18_new 14 from AlexNet import AlexNet 15 from Vgg16 import VGGNet16 16 from Densenet import DenseNet121, DenseNet169, DenseNet201, DenseNet264 17 18 from NIN import NIN_Net 19 from GoogleNet import GoogLeNet 20 from MobileNet_v3 import mobilenet_v3 21 from shuffleNet import shuffleNet_g3_ 22 23 def main(): 24 25 print("Load datasets...") 26 27 # transforms.RandomHorizontalFlip(p=0.5)---以0.5的概率对图片做水平横向翻转 28 # transforms.ToTensor()---shape从(H,W,C)->(C,H,W), 每个像素点从(0-255)映射到(0-1):直接除以255 29 # transforms.Normalize---先将输入归一化到(0,1),像素点通过"(x-mean)/std",将每个元素分布到(-1,1) 30 transform_train = transforms.Compose([ 31 transforms.Resize((224, 224), interpolation=InterpolationMode.BICUBIC), 32 # transforms.RandomCrop(32, padding=4), # 先四周填充0，在吧图像随机裁剪成32*32 33 transforms.RandomHorizontalFlip(p=0.5), 34 transforms.ToTensor(), 35 transforms.Normalize((0.485, 0.456, 0.406), (0.229, 0.224, 0.225)) 36 ]) 37 38 transform_test = transforms.Compose([ 39 transforms.Resize((224, 224), interpolation=InterpolationMode.BICUBIC), 40 # transforms.RandomCrop(32, padding=4), # 先四周填充0，在吧图像随机裁剪成32*32 41 transforms.ToTensor(), 42 transforms.Normalize((0.485, 0.456, 0.406), (0.229, 0.224, 0.225)) 43 ]) 44 45 # 内置函数下载数据集 46 train_dataset = datasets.CIFAR10(root="./data/Cifar10/", train=True, 47 transform = transform_train, 48 download=True) 49 test_dataset = datasets.CIFAR10(root = "./data/Cifar10/", 50 train = False, 51 transform = transform_test, 52 download=True) 53 54 print(len(train_dataset), len(test_dataset)) 55 56 Batch_size = 64 57 train_loader = DataLoader(train_dataset, batch_size=Batch_size, shuffle = True, num_workers=4) 58 test_loader = DataLoader(test_dataset, batch_size = Batch_size, shuffle = False, num_workers=4) 59 60 # 设置CUDA 61 device = torch.device("cuda:0" if torch.cuda.is_available() else "cpu") 62 63 # 初始化模型 64 # 直接更换模型就行，其他无需操作 65 # model = Lenet5_new().to(device) 66 # model = ResNet18().to(device) 67 # model = ResNet18_new().to(device) 68 # model = VGGNet16().to(device) 69 # model = DenseNet121().to(device) 70 # model = DenseNet169().to(device) 71 72 # model = NIN_Net().to(device) 73 74 # model = GoogLeNet().to(device) 75 # model = mobilenet_v3().to(device) 76 model = shuffleNet_g3_().to(device) 77 # model = AlexNet(num_classes=10, init_weights=True).to(device) 78 print(" shuffleNet_g3_ train...") 79 80 # 构造损失函数和优化器 81 criterion = nn.CrossEntropyLoss() # 多分类softmax构造损失 82 # opt = optim.SGD(model.parameters(), lr=0.01, momentum=0.8, weight_decay=0.001) 83 opt = optim.SGD(model.parameters(), lr=0.01, momentum=0.9, weight_decay=0.0005) 84 85 # 动态更新学习率 ------每隔step_size : lr = lr * gamma 86 schedule = optim.lr_scheduler.StepLR(opt, step_size=10, gamma=0.6, last_epoch=-1) 87 88 # 开始训练 89 print("Start Train...") 90 91 epochs = 100 92 93 loss_list = [] 94 train_acc_list =[] 95 test_acc_list = [] 96 epochs_list = [] 97 98 for epoch in range(0, epochs): 99 100 start = time.time() 101 102 model.train() 103 104 running_loss = 0.0 105 batch_num = 0 106 107 for i, (inputs, labels) in enumerate(train_loader): 108 109 inputs, labels = inputs.to(device), labels.to(device) 110 111 # 将数据送入模型训练 112 outputs = model(inputs) 113 # 计算损失 114 loss = criterion(outputs, labels).to(device) 115 116 # 重置梯度 117 opt.zero_grad() 118 # 计算梯度，反向传播 119 loss.backward() 120 # 根据反向传播的梯度值优化更新参数 121 opt.step() 122 123 # 100个batch的 loss 之和 124 running_loss += loss.item() 125 # loss_list.append(loss.item()) 126 batch_num+=1 127 128 129 epochs_list.append(epoch) 130 131 # 每一轮结束输出一下当前的学习率 lr 132 lr_1 = opt.param_groups[0]['lr'] 133 print("learn_rate:%.15f" % lr_1) 134 schedule.step() 135 136 end = time.time() 137 print('epoch = %d/100, batch_num = %d, loss = %.6f, time = %.3f' % (epoch+1, batch_num, running_loss/batch_num, end-start)) 138 running_loss=0.0 139 140 # 每个epoch训练结束，都进行一次测试验证 141 model.eval() 142 train_correct = 0.0 143 train_total = 0 144 145 test_correct = 0.0 146 test_total = 0 147 148 # 训练模式不需要反向传播更新梯度 149 with torch.no_grad(): 150 151 # print("=======================train=======================") 152 for inputs, labels in train_loader: 153 inputs, labels = inputs.to(device), labels.to(device) 154 outputs = model(inputs) 155 156 pred = outputs.argmax(dim=1) # 返回每一行中最大值元素索引 157 train_total += inputs.size(0) 158 train_correct += torch.eq(pred, labels).sum().item() 159 160 161 # print("=======================test=======================") 162 for inputs, labels in test_loader: 163 inputs, labels = inputs.to(device), labels.to(device) 164 outputs = model(inputs) 165 166 pred = outputs.argmax(dim=1) # 返回每一行中最大值元素索引 167 test_total += inputs.size(0) 168 test_correct += torch.eq(pred, labels).sum().item() 169 170 print("train_total = %d, Accuracy = %.5f %%, test_total= %d, Accuracy = %.5f %%" %(train_total, 100 * train_correct / train_total, test_total, 100 * test_correct / test_total)) 171 172 train_acc_list.append(100 * train_correct / train_total) 173 test_acc_list.append(100 * test_correct / test_total) 174 175 # print("Accuracy of the network on the 10000 test images:%.5f %%" % (100 * test_correct / test_total)) 176 # print("===============================================") 177 178 fig = plt.figure(figsize=(4, 4)) 179 180 plt.plot(epochs_list, train_acc_list, label='train_acc_list') 181 plt.plot(epochs_list, test_acc_list, label='test_acc_list') 182 plt.legend() 183 plt.title("train_test_acc") 184 plt.savefig('shuffleNet_g3_acc_epoch_{:04d}.png'.format(epochs)) 185 plt.close() 186 187 if __name__ == "__main__": 188 189 main()
loss 和acc
1 (pytorch-CycleGAN-and-pix2pix) python classfyNet_train.py 2 torch.Size([64, 10]) 3 PyTorch Version: 1.12.1+cu102 4 Torchvision Version: 0.13.1+cu102 5 Load datasets... 6 Files already downloaded and verified 7 Files already downloaded and verified 8 50000 10000 9 shuffleNet_g3_ train... 10 Start Train... 11 learn_rate:0.010000000000000 12 epoch = 1/100, batch_num = 782, loss = 2.462858, time = 64.375 13 train_total = 50000, Accuracy = 56.65400 %, test_total= 10000, Accuracy = 55.78000 % 14 learn_rate:0.010000000000000 15 epoch = 2/100, batch_num = 782, loss = 1.160206, time = 63.712 16 train_total = 50000, Accuracy = 65.57400 %, test_total= 10000, Accuracy = 63.73000 % 17 learn_rate:0.010000000000000 18 epoch = 3/100, batch_num = 782, loss = 0.939427, time = 63.848 19 train_total = 50000, Accuracy = 70.27200 %, test_total= 10000, Accuracy = 67.48000 % 20 learn_rate:0.010000000000000 21 epoch = 4/100, batch_num = 782, loss = 0.829697, time = 64.088 22 train_total = 50000, Accuracy = 73.54000 %, test_total= 10000, Accuracy = 71.55000 % 23 learn_rate:0.010000000000000 24 epoch = 5/100, batch_num = 782, loss = 0.705321, time = 63.891 25 train_total = 50000, Accuracy = 77.58000 %, test_total= 10000, Accuracy = 75.42000 % 26 learn_rate:0.010000000000000 27 epoch = 6/100, batch_num = 782, loss = 0.615606, time = 63.788 28 train_total = 50000, Accuracy = 81.87800 %, test_total= 10000, Accuracy = 78.72000 % 29 learn_rate:0.010000000000000 30 epoch = 7/100, batch_num = 782, loss = 0.558104, time = 64.142 31 train_total = 50000, Accuracy = 83.93200 %, test_total= 10000, Accuracy = 80.47000 % 32 learn_rate:0.010000000000000 33 epoch = 8/100, batch_num = 782, loss = 0.506482, time = 64.006 34 train_total = 50000, Accuracy = 86.27000 %, test_total= 10000, Accuracy = 82.37000 % 35 learn_rate:0.010000000000000 36 epoch = 9/100, batch_num = 782, loss = 0.465115, time = 63.775 37 train_total = 50000, Accuracy = 86.68000 %, test_total= 10000, Accuracy = 82.64000 % 38 learn_rate:0.010000000000000 39 epoch = 10/100, batch_num = 782, loss = 0.431530, time = 63.784 40 train_total = 50000, Accuracy = 86.92400 %, test_total= 10000, Accuracy = 82.38000 % 41 learn_rate:0.006000000000000 42 epoch = 11/100, batch_num = 782, loss = 0.344329, time = 64.106 43 train_total = 50000, Accuracy = 91.15600 %, test_total= 10000, Accuracy = 85.26000 % 44 learn_rate:0.006000000000000 45 epoch = 12/100, batch_num = 782, loss = 0.321713, time = 64.323 46 train_total = 50000, Accuracy = 91.30200 %, test_total= 10000, Accuracy = 84.90000 % 47 learn_rate:0.006000000000000 48 epoch = 13/100, batch_num = 782, loss = 0.299179, time = 63.639 49 train_total = 50000, Accuracy = 91.61200 %, test_total= 10000, Accuracy = 84.90000 % 50 learn_rate:0.006000000000000 51 epoch = 14/100, batch_num = 782, loss = 0.282726, time = 63.733 52 train_total = 50000, Accuracy = 92.61600 %, test_total= 10000, Accuracy = 85.92000 % 53 learn_rate:0.006000000000000 54 epoch = 15/100, batch_num = 782, loss = 0.269207, time = 63.728 55 train_total = 50000, Accuracy = 93.07800 %, test_total= 10000, Accuracy = 85.78000 % 56 learn_rate:0.006000000000000 57 epoch = 16/100, batch_num = 782, loss = 0.258053, time = 63.725 58 train_total = 50000, Accuracy = 93.90600 %, test_total= 10000, Accuracy = 86.29000 % 59 learn_rate:0.006000000000000 60 epoch = 17/100, batch_num = 782, loss = 0.247075, time = 63.684 61 train_total = 50000, Accuracy = 92.80400 %, test_total= 10000, Accuracy = 85.27000 % 62 learn_rate:0.006000000000000 63 epoch = 18/100, batch_num = 782, loss = 0.238726, time = 63.705 64 train_total = 50000, Accuracy = 93.13800 %, test_total= 10000, Accuracy = 85.25000 % 65 learn_rate:0.006000000000000 66 epoch = 19/100, batch_num = 782, loss = 0.220595, time = 63.783 67 train_total = 50000, Accuracy = 93.76400 %, test_total= 10000, Accuracy = 85.60000 % 68 learn_rate:0.006000000000000 69 epoch = 20/100, batch_num = 782, loss = 0.218569, time = 63.865 70 train_total = 50000, Accuracy = 93.74800 %, test_total= 10000, Accuracy = 85.93000 % 71 learn_rate:0.003600000000000 72 epoch = 21/100, batch_num = 782, loss = 0.149462, time = 63.736 73 train_total = 50000, Accuracy = 97.29800 %, test_total= 10000, Accuracy = 87.88000 % 74 learn_rate:0.003600000000000 75 epoch = 22/100, batch_num = 782, loss = 0.119382, time = 63.730 76 train_total = 50000, Accuracy = 97.49400 %, test_total= 10000, Accuracy = 87.68000 % 77 learn_rate:0.003600000000000 78 epoch = 23/100, batch_num = 782, loss = 0.109001, time = 64.248 79 train_total = 50000, Accuracy = 97.75200 %, test_total= 10000, Accuracy = 87.76000 % 80 learn_rate:0.003600000000000 81 epoch = 24/100, batch_num = 782, loss = 0.109486, time = 64.147 82 train_total = 50000, Accuracy = 97.35000 %, test_total= 10000, Accuracy = 86.76000 % 83 learn_rate:0.003600000000000 84 epoch = 25/100, batch_num = 782, loss = 0.109683, time = 63.822 85 train_total = 50000, Accuracy = 98.13000 %, test_total= 10000, Accuracy = 87.84000 % 86 learn_rate:0.003600000000000 87 epoch = 26/100, batch_num = 782, loss = 0.098647, time = 63.886 88 train_total = 50000, Accuracy = 98.16800 %, test_total= 10000, Accuracy = 87.58000 % 89 learn_rate:0.003600000000000 90 epoch = 27/100, batch_num = 782, loss = 0.101305, time = 63.814 91 train_total = 50000, Accuracy = 97.82600 %, test_total= 10000, Accuracy = 87.36000 % 92 learn_rate:0.003600000000000 93 epoch = 28/100, batch_num = 782, loss = 0.099845, time = 63.648 94 train_total = 50000, Accuracy = 98.02600 %, test_total= 10000, Accuracy = 87.77000 % 95 learn_rate:0.003600000000000 96 epoch = 29/100, batch_num = 782, loss = 0.091574, time = 63.675 97 train_total = 50000, Accuracy = 98.26000 %, test_total= 10000, Accuracy = 87.52000 % 98 learn_rate:0.003600000000000 99 epoch = 30/100, batch_num = 782, loss = 0.091463, time = 64.017 100 train_total = 50000, Accuracy = 98.21800 %, test_total= 10000, Accuracy = 86.83000 % 101 learn_rate:0.002160000000000 102 epoch = 31/100, batch_num = 782, loss = 0.057245, time = 63.766 103 train_total = 50000, Accuracy = 99.53600 %, test_total= 10000, Accuracy = 88.44000 % 104 learn_rate:0.002160000000000 105 epoch = 32/100, batch_num = 782, loss = 0.039178, time = 63.689 106 train_total = 50000, Accuracy = 99.72000 %, test_total= 10000, Accuracy = 88.72000 % 107 learn_rate:0.002160000000000 108 epoch = 33/100, batch_num = 782, loss = 0.030545, time = 63.836 109 train_total = 50000, Accuracy = 99.83000 %, test_total= 10000, Accuracy = 88.49000 % 110 learn_rate:0.002160000000000 111 epoch = 34/100, batch_num = 782, loss = 0.026627, time = 63.690 112 train_total = 50000, Accuracy = 99.83200 %, test_total= 10000, Accuracy = 88.66000 % 113 learn_rate:0.002160000000000 114 epoch = 35/100, batch_num = 782, loss = 0.024787, time = 63.707 115 train_total = 50000, Accuracy = 99.88600 %, test_total= 10000, Accuracy = 88.74000 % 116 learn_rate:0.002160000000000 117 epoch = 36/100, batch_num = 782, loss = 0.021373, time = 63.808 118 train_total = 50000, Accuracy = 99.92400 %, test_total= 10000, Accuracy = 88.69000 % 119 learn_rate:0.002160000000000 120 epoch = 37/100, batch_num = 782, loss = 0.022256, time = 63.757 121 train_total = 50000, Accuracy = 99.93200 %, test_total= 10000, Accuracy = 88.60000 % 122 learn_rate:0.002160000000000 123 epoch = 38/100, batch_num = 782, loss = 0.019585, time = 64.066 124 train_total = 50000, Accuracy = 99.85400 %, test_total= 10000, Accuracy = 88.53000 % 125 learn_rate:0.002160000000000 126 epoch = 39/100, batch_num = 782, loss = 0.018379, time = 63.657 127 train_total = 50000, Accuracy = 99.89000 %, test_total= 10000, Accuracy = 88.72000 % 128 learn_rate:0.002160000000000 129 epoch = 40/100, batch_num = 782, loss = 0.021633, time = 63.823 130 train_total = 50000, Accuracy = 99.93200 %, test_total= 10000, Accuracy = 88.59000 % 131 learn_rate:0.001296000000000 132 epoch = 41/100, batch_num = 782, loss = 0.013789, time = 64.312 133 train_total = 50000, Accuracy = 99.99400 %, test_total= 10000, Accuracy = 89.13000 % 134 learn_rate:0.001296000000000 135 epoch = 42/100, batch_num = 782, loss = 0.010126, time = 63.708 136 train_total = 50000, Accuracy = 99.99200 %, test_total= 10000, Accuracy = 89.23000 % 137 learn_rate:0.001296000000000 138 epoch = 43/100, batch_num = 782, loss = 0.009423, time = 63.673 139 train_total = 50000, Accuracy = 99.99000 %, test_total= 10000, Accuracy = 89.28000 % 140 learn_rate:0.001296000000000 141 epoch = 44/100, batch_num = 782, loss = 0.007520, time = 63.867 142 train_total = 50000, Accuracy = 99.99600 %, test_total= 10000, Accuracy = 89.22000 % 143 learn_rate:0.001296000000000 144 epoch = 45/100, batch_num = 782, loss = 0.008270, time = 63.773 145 train_total = 50000, Accuracy = 99.99400 %, test_total= 10000, Accuracy = 89.32000 % 146 learn_rate:0.001296000000000 147 epoch = 46/100, batch_num = 782, loss = 0.008582, time = 63.805 148 train_total = 50000, Accuracy = 99.99800 %, test_total= 10000, Accuracy = 89.30000 % 149 learn_rate:0.001296000000000 150 epoch = 47/100, batch_num = 782, loss = 0.006985, time = 63.897 151 train_total = 50000, Accuracy = 99.99600 %, test_total= 10000, Accuracy = 89.15000 % 152 learn_rate:0.001296000000000 153 epoch = 48/100, batch_num = 782, loss = 0.007605, time = 63.940 154 train_total = 50000, Accuracy = 99.99800 %, test_total= 10000, Accuracy = 89.28000 % 155 learn_rate:0.001296000000000 156 epoch = 49/100, batch_num = 782, loss = 0.005922, time = 64.020 157 train_total = 50000, Accuracy = 100.00000 %, test_total= 10000, Accuracy = 89.11000 % 158 learn_rate:0.001296000000000 159 epoch = 50/100, batch_num = 782, loss = 0.005957, time = 63.700 160 train_total = 50000, Accuracy = 100.00000 %, test_total= 10000, Accuracy = 89.16000 % 161 learn_rate:0.000777600000000 162 epoch = 51/100, batch_num = 782, loss = 0.005145, time = 63.793 163 train_total = 50000, Accuracy = 99.99800 %, test_total= 10000, Accuracy = 89.23000 % 164 learn_rate:0.000777600000000 165 epoch = 52/100, batch_num = 782, loss = 0.005367, time = 63.921 166 train_total = 50000, Accuracy = 100.00000 %, test_total= 10000, Accuracy = 89.28000 % 167 learn_rate:0.000777600000000 168 epoch = 53/100, batch_num = 782, loss = 0.005168, time = 64.264 169 train_total = 50000, Accuracy = 100.00000 %, test_total= 10000, Accuracy = 89.28000 % 170 learn_rate:0.000777600000000 171 epoch = 54/100, batch_num = 782, loss = 0.005301, time = 64.002 172 train_total = 50000, Accuracy = 100.00000 %, test_total= 10000, Accuracy = 89.19000 % 173 learn_rate:0.000777600000000 174 epoch = 55/100, batch_num = 782, loss = 0.004463, time = 63.766 175 train_total = 50000, Accuracy = 100.00000 %, test_total= 10000, Accuracy = 89.18000 % 176 learn_rate:0.000777600000000 177 epoch = 56/100, batch_num = 782, loss = 0.004110, time = 63.826 178 train_total = 50000, Accuracy = 100.00000 %, test_total= 10000, Accuracy = 89.29000 % 179 learn_rate:0.000777600000000 180 epoch = 57/100, batch_num = 782, loss = 0.004242, time = 63.747 181 train_total = 50000, Accuracy = 100.00000 %, test_total= 10000, Accuracy = 89.27000 % 182 learn_rate:0.000777600000000 183 epoch = 58/100, batch_num = 782, loss = 0.004406, time = 63.867 184 train_total = 50000, Accuracy = 100.00000 %, test_total= 10000, Accuracy = 89.23000 % 185 learn_rate:0.000777600000000 186 epoch = 59/100, batch_num = 782, loss = 0.004135, time = 63.828 187 train_total = 50000, Accuracy = 99.99800 %, test_total= 10000, Accuracy = 89.34000 % 188 learn_rate:0.000777600000000 189 epoch = 60/100, batch_num = 782, loss = 0.004014, time = 63.833 190 train_total = 50000, Accuracy = 100.00000 %, test_total= 10000, Accuracy = 89.37000 % 191 learn_rate:0.000466560000000 192 epoch = 61/100, batch_num = 782, loss = 0.003722, time = 63.610 193 train_total = 50000, Accuracy = 100.00000 %, test_total= 10000, Accuracy = 89.45000 % 194 learn_rate:0.000466560000000 195 epoch = 62/100, batch_num = 782, loss = 0.003552, time = 63.660 196 train_total = 50000, Accuracy = 100.00000 %, test_total= 10000, Accuracy = 89.36000 % 197 learn_rate:0.000466560000000 198 epoch = 63/100, batch_num = 782, loss = 0.003629, time = 64.358 199 train_total = 50000, Accuracy = 100.00000 %, test_total= 10000, Accuracy = 89.33000 % 200 learn_rate:0.000466560000000 201 epoch = 64/100, batch_num = 782, loss = 0.003581, time = 63.670 202 train_total = 50000, Accuracy = 100.00000 %, test_total= 10000, Accuracy = 89.44000 % 203 learn_rate:0.000466560000000 204 epoch = 65/100, batch_num = 782, loss = 0.003261, time = 64.103 205 train_total = 50000, Accuracy = 100.00000 %, test_total= 10000, Accuracy = 89.29000 % 206 learn_rate:0.000466560000000 207 epoch = 66/100, batch_num = 782, loss = 0.003679, time = 63.685 208 train_total = 50000, Accuracy = 100.00000 %, test_total= 10000, Accuracy = 89.31000 % 209 learn_rate:0.000466560000000 210 epoch = 67/100, batch_num = 782, loss = 0.003586, time = 64.044 211 train_total = 50000, Accuracy = 100.00000 %, test_total= 10000, Accuracy = 89.17000 % 212 learn_rate:0.000466560000000 213 epoch = 68/100, batch_num = 782, loss = 0.003351, time = 64.318 214 train_total = 50000, Accuracy = 100.00000 %, test_total= 10000, Accuracy = 89.52000 % 215 learn_rate:0.000466560000000 216 epoch = 69/100, batch_num = 782, loss = 0.003492, time = 63.885 217 train_total = 50000, Accuracy = 100.00000 %, test_total= 10000, Accuracy = 89.40000 % 218 learn_rate:0.000466560000000 219 epoch = 70/100, batch_num = 782, loss = 0.003522, time = 63.692 220 train_total = 50000, Accuracy = 100.00000 %, test_total= 10000, Accuracy = 89.27000 % 221 learn_rate:0.000279936000000 222 epoch = 71/100, batch_num = 782, loss = 0.003545, time = 64.041 223 train_total = 50000, Accuracy = 100.00000 %, test_total= 10000, Accuracy = 89.27000 % 224 learn_rate:0.000279936000000 225 epoch = 72/100, batch_num = 782, loss = 0.003211, time = 64.332 226 train_total = 50000, Accuracy = 100.00000 %, test_total= 10000, Accuracy = 89.40000 % 227 learn_rate:0.000279936000000 228 epoch = 73/100, batch_num = 782, loss = 0.003316, time = 63.811 229 train_total = 50000, Accuracy = 100.00000 %, test_total= 10000, Accuracy = 89.44000 % 230 learn_rate:0.000279936000000 231 epoch = 74/100, batch_num = 782, loss = 0.003299, time = 64.060 232 train_total = 50000, Accuracy = 100.00000 %, test_total= 10000, Accuracy = 89.45000 % 233 learn_rate:0.000279936000000 234 epoch = 75/100, batch_num = 782, loss = 0.003383, time = 63.998 235 train_total = 50000, Accuracy = 100.00000 %, test_total= 10000, Accuracy = 89.33000 % 236 learn_rate:0.000279936000000 237 epoch = 76/100, batch_num = 782, loss = 0.003282, time = 63.747 238 train_total = 50000, Accuracy = 100.00000 %, test_total= 10000, Accuracy = 89.29000 % 239 learn_rate:0.000279936000000 240 epoch = 77/100, batch_num = 782, loss = 0.002967, time = 63.960 241 train_total = 50000, Accuracy = 100.00000 %, test_total= 10000, Accuracy = 89.18000 % 242 learn_rate:0.000279936000000 243 epoch = 78/100, batch_num = 782, loss = 0.003005, time = 63.686 244 train_total = 50000, Accuracy = 100.00000 %, test_total= 10000, Accuracy = 89.34000 % 245 learn_rate:0.000279936000000 246 epoch = 79/100, batch_num = 782, loss = 0.003053, time = 63.720 247 train_total = 50000, Accuracy = 100.00000 %, test_total= 10000, Accuracy = 89.32000 % 248 learn_rate:0.000279936000000 249 epoch = 80/100, batch_num = 782, loss = 0.003222, time = 63.766 250 train_total = 50000, Accuracy = 100.00000 %, test_total= 10000, Accuracy = 89.42000 % 251 learn_rate:0.000167961600000 252 epoch = 81/100, batch_num = 782, loss = 0.002952, time = 63.796 253 train_total = 50000, Accuracy = 100.00000 %, test_total= 10000, Accuracy = 89.21000 % 254 learn_rate:0.000167961600000 255 epoch = 82/100, batch_num = 782, loss = 0.003042, time = 63.742 256 train_total = 50000, Accuracy = 100.00000 %, test_total= 10000, Accuracy = 89.34000 % 257 learn_rate:0.000167961600000 258 epoch = 83/100, batch_num = 782, loss = 0.003519, time = 63.730 259 train_total = 50000, Accuracy = 100.00000 %, test_total= 10000, Accuracy = 89.33000 % 260 learn_rate:0.000167961600000 261 epoch = 84/100, batch_num = 782, loss = 0.003047, time = 64.059 262 train_total = 50000, Accuracy = 100.00000 %, test_total= 10000, Accuracy = 89.25000 % 263 learn_rate:0.000167961600000 264 epoch = 85/100, batch_num = 782, loss = 0.002967, time = 63.633 265 train_total = 50000, Accuracy = 100.00000 %, test_total= 10000, Accuracy = 89.51000 % 266 learn_rate:0.000167961600000 267 epoch = 86/100, batch_num = 782, loss = 0.002872, time = 64.210 268 train_total = 50000, Accuracy = 100.00000 %, test_total= 10000, Accuracy = 89.56000 % 269 learn_rate:0.000167961600000 270 epoch = 87/100, batch_num = 782, loss = 0.003119, time = 64.020 271 train_total = 50000, Accuracy = 100.00000 %, test_total= 10000, Accuracy = 89.21000 % 272 learn_rate:0.000167961600000 273 epoch = 88/100, batch_num = 782, loss = 0.002836, time = 64.013 274 train_total = 50000, Accuracy = 100.00000 %, test_total= 10000, Accuracy = 89.36000 % 275 learn_rate:0.000167961600000 276 epoch = 89/100, batch_num = 782, loss = 0.002810, time = 64.153 277 train_total = 50000, Accuracy = 100.00000 %, test_total= 10000, Accuracy = 89.26000 % 278 learn_rate:0.000167961600000 279 epoch = 90/100, batch_num = 782, loss = 0.002952, time = 64.026 280 train_total = 50000, Accuracy = 100.00000 %, test_total= 10000, Accuracy = 89.29000 % 281 learn_rate:0.000100776960000 282 epoch = 91/100, batch_num = 782, loss = 0.002950, time = 63.881 283 train_total = 50000, Accuracy = 100.00000 %, test_total= 10000, Accuracy = 89.35000 % 284 learn_rate:0.000100776960000 285 epoch = 92/100, batch_num = 782, loss = 0.002839, time = 65.006 286 train_total = 50000, Accuracy = 100.00000 %, test_total= 10000, Accuracy = 89.40000 % 287 learn_rate:0.000100776960000 288 epoch = 93/100, batch_num = 782, loss = 0.002889, time = 65.473 289 train_total = 50000, Accuracy = 100.00000 %, test_total= 10000, Accuracy = 89.34000 % 290 learn_rate:0.000100776960000 291 epoch = 94/100, batch_num = 782, loss = 0.002764, time = 64.265 292 train_total = 50000, Accuracy = 100.00000 %, test_total= 10000, Accuracy = 89.30000 % 293 learn_rate:0.000100776960000 294 epoch = 95/100, batch_num = 782, loss = 0.003014, time = 63.756 295 train_total = 50000, Accuracy = 100.00000 %, test_total= 10000, Accuracy = 89.25000 % 296 learn_rate:0.000100776960000 297 epoch = 96/100, batch_num = 782, loss = 0.002896, time = 63.826 298 train_total = 50000, Accuracy = 100.00000 %, test_total= 10000, Accuracy = 89.26000 % 299 learn_rate:0.000100776960000 300 epoch = 97/100, batch_num = 782, loss = 0.002860, time = 63.741 301 train_total = 50000, Accuracy = 100.00000 %, test_total= 10000, Accuracy = 89.39000 % 302 learn_rate:0.000100776960000 303 epoch = 98/100, batch_num = 782, loss = 0.002967, time = 63.768 304 train_total = 50000, Accuracy = 100.00000 %, test_total= 10000, Accuracy = 89.31000 % 305 learn_rate:0.000100776960000 306 epoch = 99/100, batch_num = 782, loss = 0.002842, time = 63.707 307 train_total = 50000, Accuracy = 100.00000 %, test_total= 10000, Accuracy = 89.30000 % 308 learn_rate:0.000100776960000 309 epoch = 100/100, batch_num = 782, loss = 0.002862, time = 63.831 310 train_total = 50000, Accuracy = 100.00000 %, test_total= 10000, Accuracy = 89.45000 %
图shuffleNet_g3_acc_epoch_100
