# 数据见Fcat_20210419# 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 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 ='/_W28_04_BJGB3_BLN_S_C10'# In[1] 加载数据# In[2]#图像预处理变换的定义
transform_train = transforms.Compose([
transforms.RandomCrop(32, padding=4),#在一个随机的位置进行裁剪,32正方形裁剪,每个边框上填充4
transforms.RandomHorizontalFlip(),#以给定的概率随机水平翻转给定的PIL图像,默认值为0.5
transforms.ToTensor(),#将PIL Image或者 ndarray 转换为tensor,并且归一化至[0-1]
transforms.Normalize((0.4914,0.4822,0.4465),(0.2023,0.1994,0.2010)),#用平均值和标准偏差归一化张量图像,#(M1,…,Mn)和(S1,…,Sn)将标准化输入的每个通道])
transform_test = transforms.Compose([#测试集同样进行图像预处理
transforms.ToTensor(),
transforms.Normalize((0.4914,0.4822,0.4465),(0.2023,0.1994,0.2010)),])
cifar10_train = torchvision.datasets.CIFAR10(root='/home/megstudio/dataset', train=True, download=False, transform=transform_train)
cifar10_test = torchvision.datasets.CIFAR10(root='/home/megstudio/dataset', train=False, download=False, transform=transform_test)
train_loader = torch.utils.data.DataLoader(cifar10_train, batch_size=batch_size, shuffle=True, num_workers=2)
test_loader = torch.utils.data.DataLoader(cifar10_test, batch_size=tbatch_size, shuffle=False, num_workers=2)# In[1] 加载模型# Code modified from https://github.com/xternalz/WideResNet-pytorchclassBasicBlock(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)orNonedefforward(self, x):ifnot 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)classNetworkBlock(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 inrange(nb_layers):
layers.append(block(i ==0and in_planes or out_planes, out_planes, i ==0and stride or1, dropRate))return nn.Sequential(*layers)defforward(self, x):return self.layer(x)classWideResNet(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():ifisinstance(m, nn.Conv2d):
n = m.kernel_size[0]* m.kernel_size[1]* m.out_channels
m.weight.data.normal_(0, math.sqrt(2./ n))elifisinstance(m, nn.BatchNorm2d):
m.weight.data.fill_(1)
m.bias.data.zero_()elifisinstance(m, nn.Linear):
m.bias.data.zero_()defforward(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)defwrn(**kwargs):"""
Constructs a Wide Residual Networks.
"""
model = WideResNet(**kwargs)return model
defsmooth_step(a,b,c,d,x):
level_s=0.01
level_m=0.1
level_n=0.01
level_r=0.005if 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defupdate_lr(optimizer, lr):for param_group in optimizer.param_groups:
param_group['lr']= lr
# In[1] 定义测试函数deftest(model,test_loader):
model.eval()with torch.no_grad():
correct =0
total =0for 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# =============================================================================defmkdir(path):
folder = os.path.exists(path)ifnot 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 *= 3print(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=10, 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# =============================================================================defsaveModel(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_ratefor epoch inrange(epoch_s, num_epochs):#scheduler.step()
in_epoch = time.time()for i,(images, labels)inenumerate(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)
cuda
/home/megstudio/workspace/__20210430/_W28_04_BJGB3_BLN_S_C10
--- There is this folder! ------ Create a new model...---
Epoch [1/200], Step [100/500] Loss:2.1790
Epoch [1/200], Step [200/500] Loss:2.1559
Epoch [1/200], Step [300/500] Loss:2.0387
Epoch [1/200], Step [400/500] Loss:2.0988
Epoch [1/200], Step [500/500] Loss:1.8831
Accuracy of the model on the test images:15.8%
use 0.0min21.123395442962646s
Epoch [2/200], Step [100/500] Loss:2.2224
Epoch [2/200], Step [200/500] Loss:2.1422
Epoch [2/200], Step [300/500] Loss:1.9208
Epoch [2/200], Step [400/500] Loss:2.0676
Epoch [2/200], Step [500/500] Loss:1.8473
Accuracy of the model on the test images:38.1%
use 0.0min20.357621431350708s
Epoch [3/200], Step [100/500] Loss:1.9768
Epoch [3/200], Step [200/500] Loss:2.0925
Epoch [3/200], Step [300/500] Loss:1.8357
Epoch [3/200], Step [400/500] Loss:1.8657
Epoch [3/200], Step [500/500] Loss:2.0923
Accuracy of the model on the test images:43.41%
use 0.0min20.361759662628174s
Epoch [4/200], Step [100/500] Loss:2.0043
Epoch [4/200], Step [200/500] Loss:1.9353
Epoch [4/200], Step [300/500] Loss:1.8797
Epoch [4/200], Step [400/500] Loss:1.6698
Epoch [4/200], Step [500/500] Loss:1.5369
Accuracy of the model on the test images:48.11%
use 0.0min20.455519914627075s
Epoch [5/200], Step [100/500] Loss:1.9625
Epoch [5/200], Step [200/500] Loss:1.8843
Epoch [5/200], Step [300/500] Loss:2.0839
Epoch [5/200], Step [400/500] Loss:1.7186
Epoch [5/200], Step [500/500] Loss:1.9012
Accuracy of the model on the test images:47.71%
use 0.0min20.377105951309204s
Epoch [6/200], Step [100/500] Loss:1.9017
Epoch [6/200], Step [200/500] Loss:1.8785
Epoch [6/200], Step [300/500] Loss:1.9705
Epoch [6/200], Step [400/500] Loss:1.6972
Epoch [6/200], Step [500/500] Loss:1.3769
Accuracy of the model on the test images:57.25%
use 0.0min20.3419349193573s
Epoch [7/200], Step [100/500] Loss:1.4513
Epoch [7/200], Step [200/500] Loss:1.5923
Epoch [7/200], Step [300/500] Loss:1.8236
Epoch [7/200], Step [400/500] Loss:1.9022
Epoch [7/200], Step [500/500] Loss:1.9974
Accuracy of the model on the test images:56.02%
use 0.0min20.339054822921753s
Epoch [8/200], Step [100/500] Loss:2.0042
Epoch [8/200], Step [200/500] Loss:1.3628
Epoch [8/200], Step [300/500] Loss:1.9355
Epoch [8/200], Step [400/500] Loss:1.2330
Epoch [8/200], Step [500/500] Loss:1.4815
Accuracy of the model on the test images:58.41%
use 0.0min20.403385639190674s
Epoch [9/200], Step [100/500] Loss:1.1941
Epoch [9/200], Step [200/500] Loss:1.1131
Epoch [9/200], Step [300/500] Loss:1.4155
Epoch [9/200], Step [400/500] Loss:1.7827
Epoch [9/200], Step [500/500] Loss:1.3642
Accuracy of the model on the test images:64.23%
use 0.0min20.375659942626953s
Epoch [10/200], Step [100/500] Loss:1.4404
Epoch [10/200], Step [200/500] Loss:1.7268
Epoch [10/200], Step [300/500] Loss:1.9433
Epoch [10/200], Step [400/500] Loss:1.4888
Epoch [10/200], Step [500/500] Loss:1.1935
Accuracy of the model on the test images:63.67%
use 0.0min20.35075068473816s
Epoch [11/200], Step [100/500] Loss:1.4947
Epoch [11/200], Step [200/500] Loss:1.7364
Epoch [11/200], Step [300/500] Loss:1.8127
Epoch [11/200], Step [400/500] Loss:1.4353
Epoch [11/200], Step [500/500] Loss:1.7878
Accuracy of the model on the test images:67.08%
use 0.0min20.35635018348694s
Epoch [12/200], Step [100/500] Loss:1.8190
Epoch [12/200], Step [200/500] Loss:1.7807
Epoch [12/200], Step [300/500] Loss:1.1506
Epoch [12/200], Step [400/500] Loss:1.6618
Epoch [12/200], Step [500/500] Loss:1.4403
Accuracy of the model on the test images:69.18%
use 0.0min20.43039631843567s
Epoch [13/200], Step [100/500] Loss:1.7314
Epoch [13/200], Step [200/500] Loss:0.9851
Epoch [13/200], Step [300/500] Loss:1.6270
Epoch [13/200], Step [400/500] Loss:1.0374
Epoch [13/200], Step [500/500] Loss:1.2373
Accuracy of the model on the test images:70.25%
use 0.0min20.34923028945923s
Epoch [14/200], Step [100/500] Loss:1.4373
Epoch [14/200], Step [200/500] Loss:1.0507
Epoch [14/200], Step [300/500] Loss:1.7159
Epoch [14/200], Step [400/500] Loss:1.6959
Epoch [14/200], Step [500/500] Loss:1.6582
Accuracy of the model on the test images:70.07%
use 0.0min20.372792959213257s
Epoch [15/200], Step [100/500] Loss:1.5673
Epoch [15/200], Step [200/500] Loss:1.6747
Epoch [15/200], Step [300/500] Loss:1.2791
Epoch [15/200], Step [400/500] Loss:1.6326
Epoch [15/200], Step [500/500] Loss:1.1707
Accuracy of the model on the test images:72.36%
use 0.0min20.39280652999878s
Epoch [16/200], Step [100/500] Loss:1.6406
Epoch [16/200], Step [200/500] Loss:1.4876
Epoch [16/200], Step [300/500] Loss:1.7292
Epoch [16/200], Step [400/500] Loss:1.6747
Epoch [16/200], Step [500/500] Loss:1.4707
Accuracy of the model on the test images:75.81%
use 0.0min20.440183401107788s
Epoch [17/200], Step [100/500] Loss:1.8210
Epoch [17/200], Step [200/500] Loss:1.2006
Epoch [17/200], Step [300/500] Loss:1.8593
Epoch [17/200], Step [400/500] Loss:1.6482
Epoch [17/200], Step [500/500] Loss:1.7049
Accuracy of the model on the test images:73.82%
use 0.0min20.38654136657715s
Epoch [18/200], Step [100/500] Loss:1.6006
Epoch [18/200], Step [200/500] Loss:1.4309
Epoch [18/200], Step [300/500] Loss:1.1157
Epoch [18/200], Step [400/500] Loss:1.2330
Epoch [18/200], Step [500/500] Loss:1.5895
Accuracy of the model on the test images:76.96%
use 0.0min20.359359979629517s
Epoch [19/200], Step [100/500] Loss:1.0131
Epoch [19/200], Step [200/500] Loss:1.5762
Epoch [19/200], Step [300/500] Loss:0.8004
Epoch [19/200], Step [400/500] Loss:1.4015
Epoch [19/200], Step [500/500] Loss:1.8083
Accuracy of the model on the test images:75.61%
use 0.0min20.411002159118652s
Epoch [20/200], Step [100/500] Loss:1.3020
Epoch [20/200], Step [200/500] Loss:1.5370
Epoch [20/200], Step [300/500] Loss:1.1880
Epoch [20/200], Step [400/500] Loss:1.5604
Epoch [20/200], Step [500/500] Loss:1.4079
Accuracy of the model on the test images:75.06%
use 0.0min20.35499906539917s
Epoch [21/200], Step [100/500] Loss:1.3176
Epoch [21/200], Step [200/500] Loss:1.5449
Epoch [21/200], Step [300/500] Loss:1.5561
Epoch [21/200], Step [400/500] Loss:1.5398
Epoch [21/200], Step [500/500] Loss:1.0853
Accuracy of the model on the test images:79.0%
use 0.0min20.335134744644165s
Epoch [22/200], Step [100/500] Loss:1.4288
Epoch [22/200], Step [200/500] Loss:1.5144
Epoch [22/200], Step [300/500] Loss:1.0419
Epoch [22/200], Step [400/500] Loss:0.6797
Epoch [22/200], Step [500/500] Loss:1.6279
Accuracy of the model on the test images:80.12%
use 0.0min20.401585578918457s
Epoch [23/200], Step [100/500] Loss:0.7904
Epoch [23/200], Step [200/500] Loss:0.9844
Epoch [23/200], Step [300/500] Loss:1.5441
Epoch [23/200], Step [400/500] Loss:1.3214
Epoch [23/200], Step [500/500] Loss:1.5232
Accuracy of the model on the test images:80.41%
use 0.0min20.373417377471924s
Epoch [24/200], Step [100/500] Loss:1.4240
Epoch [24/200], Step [200/500] Loss:1.4639
Epoch [24/200], Step [300/500] Loss:1.5579
Epoch [24/200], Step [400/500] Loss:1.7012
Epoch [24/200], Step [500/500] Loss:1.1945
Accuracy of the model on the test images:81.64%
use 0.0min20.36199688911438s
Epoch [25/200], Step [100/500] Loss:0.7147
Epoch [25/200], Step [200/500] Loss:1.7029
Epoch [25/200], Step [300/500] Loss:0.7409
Epoch [25/200], Step [400/500] Loss:1.4500
Epoch [25/200], Step [500/500] Loss:1.2520
Accuracy of the model on the test images:80.97%
use 0.0min20.388251781463623s
Epoch [26/200], Step [100/500] Loss:1.4758
Epoch [26/200], Step [200/500] Loss:1.3249
Epoch [26/200], Step [300/500] Loss:0.9811
Epoch [26/200], Step [400/500] Loss:1.7301
Epoch [26/200], Step [500/500] Loss:1.6176
Accuracy of the model on the test images:83.47%
use 0.0min20.412151336669922s
Epoch [27/200], Step [100/500] Loss:1.3818
Epoch [27/200], Step [200/500] Loss:1.2675
Epoch [27/200], Step [300/500] Loss:1.0948
Epoch [27/200], Step [400/500] Loss:1.3120
Epoch [27/200], Step [500/500] Loss:1.4844
Accuracy of the model on the test images:83.34%
use 0.0min20.373883962631226s
Epoch [28/200], Step [100/500] Loss:0.6394
Epoch [28/200], Step [200/500] Loss:1.4804
Epoch [28/200], Step [300/500] Loss:1.1436
Epoch [28/200], Step [400/500] Loss:1.4518
Epoch [28/200], Step [500/500] Loss:1.5433
Accuracy of the model on the test images:84.37%
use 0.0min20.37519359588623s
Epoch [29/200], Step [100/500] Loss:1.4559
Epoch [29/200], Step [200/500] Loss:1.6367
Epoch [29/200], Step [300/500] Loss:1.2694
Epoch [29/200], Step [400/500] Loss:1.4473
Epoch [29/200], Step [500/500] Loss:1.1265
Accuracy of the model on the test images:84.67%
use 0.0min20.359278917312622s
Epoch [30/200], Step [100/500] Loss:1.3819
Epoch [30/200], Step [200/500] Loss:0.5442
Epoch [30/200], Step [300/500] Loss:1.6324
Epoch [30/200], Step [400/500] Loss:1.5540
Epoch [30/200], Step [500/500] Loss:1.3784
Accuracy of the model on the test images:85.49%
use 0.0min20.419134616851807s
Epoch [31/200], Step [100/500] Loss:0.7779
Epoch [31/200], Step [200/500] Loss:0.9519
Epoch [31/200], Step [300/500] Loss:1.1823
Epoch [31/200], Step [400/500] Loss:1.4914
Epoch [31/200], Step [500/500] Loss:1.0552
Accuracy of the model on the test images:85.39%
use 0.0min20.369332313537598s
Epoch [32/200], Step [100/500] Loss:0.3527
Epoch [32/200], Step [200/500] Loss:0.5136
Epoch [32/200], Step [300/500] Loss:1.1930
Epoch [32/200], Step [400/500] Loss:0.5148
Epoch [32/200], Step [500/500] Loss:0.6566
Accuracy of the model on the test images:86.03%
use 0.0min20.37297773361206s
Epoch [33/200], Step [100/500] Loss:0.5352
Epoch [33/200], Step [200/500] Loss:1.0109
Epoch [33/200], Step [300/500] Loss:0.4268
Epoch [33/200], Step [400/500] Loss:1.0549
Epoch [33/200], Step [500/500] Loss:1.2225
Accuracy of the model on the test images:85.99%
use 0.0min20.465530157089233s
Epoch [34/200], Step [100/500] Loss:1.2413
Epoch [34/200], Step [200/500] Loss:1.3109
Epoch [34/200], Step [300/500] Loss:1.0909
Epoch [34/200], Step [400/500] Loss:0.9841
Epoch [34/200], Step [500/500] Loss:1.1050
Accuracy of the model on the test images:87.44%
use 0.0min20.359282970428467s
Epoch [35/200], Step [100/500] Loss:0.5035
Epoch [35/200], Step [200/500] Loss:1.1636
Epoch [35/200], Step [300/500] Loss:1.4892
Epoch [35/200], Step [400/500] Loss:1.3404
Epoch [35/200], Step [500/500] Loss:0.9692
Accuracy of the model on the test images:86.94%
use 0.0min20.34918475151062s
Epoch [36/200], Step [100/500] Loss:0.2755
Epoch [36/200], Step [200/500] Loss:1.3625
Epoch [36/200], Step [300/500] Loss:0.8354
Epoch [36/200], Step [400/500] Loss:1.3214
Epoch [36/200], Step [500/500] Loss:0.5478
Accuracy of the model on the test images:87.78%
use 0.0min20.52253794670105s
Epoch [37/200], Step [100/500] Loss:1.2926
Epoch [37/200], Step [200/500] Loss:0.9581
Epoch [37/200], Step [300/500] Loss:0.6451
Epoch [37/200], Step [400/500] Loss:1.2414
Epoch [37/200], Step [500/500] Loss:1.0961
Accuracy of the model on the test images:87.79%
use 0.0min20.383507251739502s
Epoch [38/200], Step [100/500] Loss:1.4111
Epoch [38/200], Step [200/500] Loss:0.4365
Epoch [38/200], Step [300/500] Loss:1.3412
Epoch [38/200], Step [400/500] Loss:1.4119
Epoch [38/200], Step [500/500] Loss:1.4039
Accuracy of the model on the test images:88.16%
use 0.0min20.35458493232727s
Epoch [39/200], Step [100/500] Loss:1.3965
Epoch [39/200], Step [200/500] Loss:0.3257
Epoch [39/200], Step [300/500] Loss:1.2711
Epoch [39/200], Step [400/500] Loss:0.9460
Epoch [39/200], Step [500/500] Loss:1.3978
Accuracy of the model on the test images:88.07%
use 0.0min20.341139316558838s
Epoch [40/200], Step [100/500] Loss:0.9849
Epoch [40/200], Step [200/500] Loss:1.1538
Epoch [40/200], Step [300/500] Loss:1.4707
Epoch [40/200], Step [400/500] Loss:1.3971
Epoch [40/200], Step [500/500] Loss:1.4379
Accuracy of the model on the test images:88.02%
use 0.0min20.370015144348145s
Epoch [41/200], Step [100/500] Loss:0.4366
Epoch [41/200], Step [200/500] Loss:0.9759
Epoch [41/200], Step [300/500] Loss:1.0825
Epoch [41/200], Step [400/500] Loss:1.0324
Epoch [41/200], Step [500/500] Loss:0.4587
Accuracy of the model on the test images:88.11%
use 0.0min20.34174919128418s
Epoch [42/200], Step [100/500] Loss:0.9650
Epoch [42/200], Step [200/500] Loss:0.4519
Epoch [42/200], Step [300/500] Loss:1.4696
Epoch [42/200], Step [400/500] Loss:1.3668
Epoch [42/200], Step [500/500] Loss:1.0172
Accuracy of the model on the test images:89.22%
use 0.0min20.347109079360962s
Epoch [43/200], Step [100/500] Loss:1.3966
Epoch [43/200], Step [200/500] Loss:1.0262
Epoch [43/200], Step [300/500] Loss:1.1712
Epoch [43/200], Step [400/500] Loss:1.3660
Epoch [43/200], Step [500/500] Loss:1.3209
Accuracy of the model on the test images:89.03%
use 0.0min20.35543131828308s
Epoch [44/200], Step [100/500] Loss:1.3771
Epoch [44/200], Step [200/500] Loss:1.2250
Epoch [44/200], Step [300/500] Loss:0.3338
Epoch [44/200], Step [400/500] Loss:1.2763
Epoch [44/200], Step [500/500] Loss:0.9907
Accuracy of the model on the test images:89.15%
use 0.0min20.35590386390686s
Epoch [45/200], Step [100/500] Loss:1.1190
Epoch [45/200], Step [200/500] Loss:1.3418
Epoch [45/200], Step [300/500] Loss:0.3929
Epoch [45/200], Step [400/500] Loss:0.7896
Epoch [45/200], Step [500/500] Loss:1.0600
Accuracy of the model on the test images:87.92%
use 0.0min20.3344624042511s
Epoch [46/200], Step [100/500] Loss:0.5885
Epoch [46/200], Step [200/500] Loss:1.2346
Epoch [46/200], Step [300/500] Loss:0.8508
Epoch [46/200], Step [400/500] Loss:1.2530
Epoch [46/200], Step [500/500] Loss:0.4935
Accuracy of the model on the test images:90.71%
use 0.0min20.371198892593384s
Epoch [47/200], Step [100/500] Loss:0.4391
Epoch [47/200], Step [200/500] Loss:0.7598
Epoch [47/200], Step [300/500] Loss:1.1973
Epoch [47/200], Step [400/500] Loss:1.3947
Epoch [47/200], Step [500/500] Loss:0.6789
Accuracy of the model on the test images:90.02%
use 0.0min20.366332054138184s
Epoch [48/200], Step [100/500] Loss:1.4140
Epoch [48/200], Step [200/500] Loss:1.2009
Epoch [48/200], Step [300/500] Loss:1.1860
Epoch [48/200], Step [400/500] Loss:0.3818
Epoch [48/200], Step [500/500] Loss:0.8759
Accuracy of the model on the test images:90.28%
use 0.0min20.336992502212524s
Epoch [49/200], Step [100/500] Loss:1.2034
Epoch [49/200], Step [200/500] Loss:1.0903
Epoch [49/200], Step [300/500] Loss:0.8990
Epoch [49/200], Step [400/500] Loss:0.4474
Epoch [49/200], Step [500/500] Loss:0.8824
Accuracy of the model on the test images:89.16%
use 0.0min20.347647428512573s
Epoch [50/200], Step [100/500] Loss:1.1515
Epoch [50/200], Step [200/500] Loss:1.2839
Epoch [50/200], Step [300/500] Loss:1.0883
Epoch [50/200], Step [400/500] Loss:0.8164
Epoch [50/200], Step [500/500] Loss:0.9372
Accuracy of the model on the test images:90.11%
use 0.0min20.36854887008667s
Epoch [51/200], Step [100/500] Loss:0.7564
Epoch [51/200], Step [200/500] Loss:1.0694
Epoch [51/200], Step [300/500] Loss:0.3627
Epoch [51/200], Step [400/500] Loss:0.4782
Epoch [51/200], Step [500/500] Loss:0.9174
Accuracy of the model on the test images:89.49%
use 0.0min20.361841440200806s
Epoch [52/200], Step [100/500] Loss:1.1215
Epoch [52/200], Step [200/500] Loss:1.0632
Epoch [52/200], Step [300/500] Loss:0.5354
Epoch [52/200], Step [400/500] Loss:1.1713
Epoch [52/200], Step [500/500] Loss:0.3436
Accuracy of the model on the test images:90.77%
use 0.0min20.35405945777893s
Epoch [53/200], Step [100/500] Loss:1.0341
Epoch [53/200], Step [200/500] Loss:1.2249
Epoch [53/200], Step [300/500] Loss:0.5486
Epoch [53/200], Step [400/500] Loss:1.1428
Epoch [53/200], Step [500/500] Loss:1.1575
Accuracy of the model on the test images:90.4%
use 0.0min20.34561586380005s
Epoch [54/200], Step [100/500] Loss:0.8552
Epoch [54/200], Step [200/500] Loss:0.5618
Epoch [54/200], Step [300/500] Loss:1.0586
Epoch [54/200], Step [400/500] Loss:0.5835
Epoch [54/200], Step [500/500] Loss:0.4571
Accuracy of the model on the test images:90.23%
use 0.0min20.375904083251953s
Epoch [55/200], Step [100/500] Loss:0.5887
Epoch [55/200], Step [200/500] Loss:1.2362
Epoch [55/200], Step [300/500] Loss:1.1553
Epoch [55/200], Step [400/500] Loss:1.2960
Epoch [55/200], Step [500/500] Loss:1.2422
Accuracy of the model on the test images:90.14%
use 0.0min20.35897397994995s
Epoch [56/200], Step [100/500] Loss:1.1849
Epoch [56/200], Step [200/500] Loss:0.8863
Epoch [56/200], Step [300/500] Loss:0.4298
Epoch [56/200], Step [400/500] Loss:1.3727
Epoch [56/200], Step [500/500] Loss:0.1565
Accuracy of the model on the test images:91.64%
use 0.0min20.358381986618042s
Epoch [57/200], Step [100/500] Loss:1.3350
Epoch [57/200], Step [200/500] Loss:0.7779
Epoch [57/200], Step [300/500] Loss:0.4617
Epoch [57/200], Step [400/500] Loss:1.2344
Epoch [57/200], Step [500/500] Loss:1.3262
Accuracy of the model on the test images:91.31%
use 0.0min20.415284633636475s
Epoch [58/200], Step [100/500] Loss:0.1896
Epoch [58/200], Step [200/500] Loss:0.6137
Epoch [58/200], Step [300/500] Loss:0.9304
Epoch [58/200], Step [400/500] Loss:1.0907
Epoch [58/200], Step [500/500] Loss:0.8966
Accuracy of the model on the test images:91.12%
use 0.0min20.36165738105774s
Epoch [59/200], Step [100/500] Loss:0.7631
Epoch [59/200], Step [200/500] Loss:0.5612
Epoch [59/200], Step [300/500] Loss:1.2073
Epoch [59/200], Step [400/500] Loss:1.1358
Epoch [59/200], Step [500/500] Loss:0.9003
Accuracy of the model on the test images:90.22%
use 0.0min20.355273485183716s
Epoch [60/200], Step [100/500] Loss:0.8119
Epoch [60/200], Step [200/500] Loss:0.8268
Epoch [60/200], Step [300/500] Loss:1.1098
Epoch [60/200], Step [400/500] Loss:0.7903
Epoch [60/200], Step [500/500] Loss:1.2007
Accuracy of the model on the test images:91.78%
use 0.0min20.364991903305054s
Epoch [61/200], Step [100/500] Loss:0.9761
Epoch [61/200], Step [200/500] Loss:0.4134
Epoch [61/200], Step [300/500] Loss:1.1588
Epoch [61/200], Step [400/500] Loss:1.1384
Epoch [61/200], Step [500/500] Loss:1.0511
Accuracy of the model on the test images:92.31%
use 0.0min20.366493940353394s
Epoch [62/200], Step [100/500] Loss:0.4448
Epoch [62/200], Step [200/500] Loss:1.1300
Epoch [62/200], Step [300/500] Loss:1.1617
Epoch [62/200], Step [400/500] Loss:0.9995
Epoch [62/200], Step [500/500] Loss:0.5588
Accuracy of the model on the test images:91.07%
use 0.0min20.35614776611328s
Epoch [63/200], Step [100/500] Loss:1.0792
Epoch [63/200], Step [200/500] Loss:1.1368
Epoch [63/200], Step [300/500] Loss:0.6460
Epoch [63/200], Step [400/500] Loss:0.9589
Epoch [63/200], Step [500/500] Loss:1.3234
Accuracy of the model on the test images:91.63%
use 0.0min20.352925300598145s
Epoch [64/200], Step [100/500] Loss:0.6526
Epoch [64/200], Step [200/500] Loss:1.2217
Epoch [64/200], Step [300/500] Loss:1.2415
Epoch [64/200], Step [400/500] Loss:1.2455
Epoch [64/200], Step [500/500] Loss:1.0209
Accuracy of the model on the test images:91.41%
use 0.0min20.417600870132446s
Epoch [65/200], Step [100/500] Loss:0.4426
Epoch [65/200], Step [200/500] Loss:1.2359
Epoch [65/200], Step [300/500] Loss:0.7901
Epoch [65/200], Step [400/500] Loss:1.1537
Epoch [65/200], Step [500/500] Loss:1.1570
Accuracy of the model on the test images:91.98%
use 0.0min20.355173349380493s
Epoch [66/200], Step [100/500] Loss:0.4872
Epoch [66/200], Step [200/500] Loss:0.6941
Epoch [66/200], Step [300/500] Loss:1.0854
Epoch [66/200], Step [400/500] Loss:0.7960
Epoch [66/200], Step [500/500] Loss:1.2412
Accuracy of the model on the test images:91.9%
use 0.0min20.3370361328125s
Epoch [67/200], Step [100/500] Loss:1.2178
Epoch [67/200], Step [200/500] Loss:1.0593
Epoch [67/200], Step [300/500] Loss:1.0975
Epoch [67/200], Step [400/500] Loss:1.0498
Epoch [67/200], Step [500/500] Loss:1.1520
Accuracy of the model on the test images:92.39%
use 0.0min20.368600130081177s
Epoch [68/200], Step [100/500] Loss:1.3849
Epoch [68/200], Step [200/500] Loss:1.1138
Epoch [68/200], Step [300/500] Loss:1.1563
Epoch [68/200], Step [400/500] Loss:1.0671
Epoch [68/200], Step [500/500] Loss:1.3540
Accuracy of the model on the test images:91.47%
use 0.0min20.388365983963013s
Epoch [69/200], Step [100/500] Loss:0.6983
Epoch [69/200], Step [200/500] Loss:0.8688
Epoch [69/200], Step [300/500] Loss:0.6467
Epoch [69/200], Step [400/500] Loss:1.0615
Epoch [69/200], Step [500/500] Loss:0.9949
Accuracy of the model on the test images:91.5%
use 0.0min20.358011722564697s
Epoch [70/200], Step [100/500] Loss:1.3060
Epoch [70/200], Step [200/500] Loss:0.2657
Epoch [70/200], Step [300/500] Loss:0.5019
Epoch [70/200], Step [400/500] Loss:1.1539
Epoch [70/200], Step [500/500] Loss:0.2488
Accuracy of the model on the test images:91.36%
use 0.0min20.365375995635986s
Epoch [71/200], Step [100/500] Loss:0.9794
Epoch [71/200], Step [200/500] Loss:1.1348
Epoch [71/200], Step [300/500] Loss:1.1828
Epoch [71/200], Step [400/500] Loss:0.8390
Epoch [71/200], Step [500/500] Loss:1.2209
Accuracy of the model on the test images:90.86%
use 0.0min20.40201234817505s
Epoch [72/200], Step [100/500] Loss:0.1715
Epoch [72/200], Step [200/500] Loss:0.4423
Epoch [72/200], Step [300/500] Loss:1.0107
Epoch [72/200], Step [400/500] Loss:1.0902
Epoch [72/200], Step [500/500] Loss:1.1116
Accuracy of the model on the test images:92.34%
use 0.0min20.396960020065308s
Epoch [73/200], Step [100/500] Loss:0.3739
Epoch [73/200], Step [200/500] Loss:1.1867
Epoch [73/200], Step [300/500] Loss:0.4535
Epoch [73/200], Step [400/500] Loss:1.1095
Epoch [73/200], Step [500/500] Loss:0.1533
Accuracy of the model on the test images:92.87%
use 0.0min20.382538557052612s
Epoch [74/200], Step [100/500] Loss:0.9620
Epoch [74/200], Step [200/500] Loss:1.1031
Epoch [74/200], Step [300/500] Loss:0.2878
Epoch [74/200], Step [400/500] Loss:1.1353
Epoch [74/200], Step [500/500] Loss:1.0970
Accuracy of the model on the test images:92.46%
use 0.0min20.36136531829834s
Epoch [75/200], Step [100/500] Loss:0.8250
Epoch [75/200], Step [200/500] Loss:0.9810
Epoch [75/200], Step [300/500] Loss:0.4182
Epoch [75/200], Step [400/500] Loss:0.9515
Epoch [75/200], Step [500/500] Loss:1.1185
Accuracy of the model on the test images:92.2%
use 0.0min20.370022773742676s
Epoch [76/200], Step [100/500] Loss:1.0507
Epoch [76/200], Step [200/500] Loss:1.2088
Epoch [76/200], Step [300/500] Loss:0.4608
Epoch [76/200], Step [400/500] Loss:0.6427
Epoch [76/200], Step [500/500] Loss:0.3101
Accuracy of the model on the test images:92.62%
use 0.0min20.362632989883423s
Epoch [77/200], Step [100/500] Loss:0.5323
Epoch [77/200], Step [200/500] Loss:0.2878
Epoch [77/200], Step [300/500] Loss:0.6650
Epoch [77/200], Step [400/500] Loss:1.0999
Epoch [77/200], Step [500/500] Loss:1.1225
Accuracy of the model on the test images:91.89%
use 0.0min20.352317571640015s
Epoch [78/200], Step [100/500] Loss:0.7871
Epoch [78/200], Step [200/500] Loss:0.7206
Epoch [78/200], Step [300/500] Loss:0.9529
Epoch [78/200], Step [400/500] Loss:0.7113
Epoch [78/200], Step [500/500] Loss:0.6871
Accuracy of the model on the test images:92.88%
use 0.0min20.35572576522827s
Epoch [79/200], Step [100/500] Loss:0.8967
Epoch [79/200], Step [200/500] Loss:0.7975
Epoch [79/200], Step [300/500] Loss:0.3521
Epoch [79/200], Step [400/500] Loss:1.1852
Epoch [79/200], Step [500/500] Loss:0.3066
Accuracy of the model on the test images:91.92%
use 0.0min20.411757230758667s
Epoch [80/200], Step [100/500] Loss:0.9884
Epoch [80/200], Step [200/500] Loss:0.7489
Epoch [80/200], Step [300/500] Loss:1.0278
Epoch [80/200], Step [400/500] Loss:0.2410
Epoch [80/200], Step [500/500] Loss:0.3059
Accuracy of the model on the test images:92.7%
use 0.0min20.342678785324097s
Epoch [81/200], Step [100/500] Loss:1.0701
Epoch [81/200], Step [200/500] Loss:0.8364
Epoch [81/200], Step [300/500] Loss:1.1095
Epoch [81/200], Step [400/500] Loss:0.6371
Epoch [81/200], Step [500/500] Loss:1.0618
Accuracy of the model on the test images:92.5%
use 0.0min20.350439310073853s
Epoch [82/200], Step [100/500] Loss:1.3469
Epoch [82/200], Step [200/500] Loss:1.1329
Epoch [82/200], Step [300/500] Loss:0.4903
Epoch [82/200], Step [400/500] Loss:0.8416
Epoch [82/200], Step [500/500] Loss:1.1430
Accuracy of the model on the test images:93.34%
use 0.0min20.346050262451172s
Epoch [83/200], Step [100/500] Loss:0.7893
Epoch [83/200], Step [200/500] Loss:0.6076
Epoch [83/200], Step [300/500] Loss:0.6015
Epoch [83/200], Step [400/500] Loss:1.0161
Epoch [83/200], Step [500/500] Loss:0.9365
Accuracy of the model on the test images:92.0%
use 0.0min20.381393909454346s
Epoch [84/200], Step [100/500] Loss:1.1401
Epoch [84/200], Step [200/500] Loss:1.0463
Epoch [84/200], Step [300/500] Loss:0.2169
Epoch [84/200], Step [400/500] Loss:0.2457
Epoch [84/200], Step [500/500] Loss:0.9423
Accuracy of the model on the test images:92.8%
use 0.0min20.357431411743164s
Epoch [85/200], Step [100/500] Loss:0.7488
Epoch [85/200], Step [200/500] Loss:0.7533
Epoch [85/200], Step [300/500] Loss:0.4501
Epoch [85/200], Step [400/500] Loss:0.4702
Epoch [85/200], Step [500/500] Loss:0.8354
Accuracy of the model on the test images:93.16%
use 0.0min20.340961694717407s
Epoch [86/200], Step [100/500] Loss:1.0408
Epoch [86/200], Step [200/500] Loss:1.0514
Epoch [86/200], Step [300/500] Loss:1.0801
Epoch [86/200], Step [400/500] Loss:0.9697
Epoch [86/200], Step [500/500] Loss:0.8542
Accuracy of the model on the test images:93.79%
use 0.0min20.4689519405365s
Epoch [87/200], Step [100/500] Loss:0.3422
Epoch [87/200], Step [200/500] Loss:1.0760
Epoch [87/200], Step [300/500] Loss:1.0802
Epoch [87/200], Step [400/500] Loss:0.5832
Epoch [87/200], Step [500/500] Loss:0.8596
Accuracy of the model on the test images:92.35%
use 0.0min20.37018585205078s
Epoch [88/200], Step [100/500] Loss:1.0215
Epoch [88/200], Step [200/500] Loss:0.5354
Epoch [88/200], Step [300/500] Loss:1.1278
Epoch [88/200], Step [400/500] Loss:0.5407
Epoch [88/200], Step [500/500] Loss:1.0884
Accuracy of the model on the test images:93.21%
use 0.0min20.390015602111816s
Epoch [89/200], Step [100/500] Loss:1.0806
Epoch [89/200], Step [200/500] Loss:0.9242
Epoch [89/200], Step [300/500] Loss:0.9808
Epoch [89/200], Step [400/500] Loss:0.9977
Epoch [89/200], Step [500/500] Loss:1.0073
Accuracy of the model on the test images:92.72%
use 0.0min20.356212854385376s
Epoch [90/200], Step [100/500] Loss:1.2423
Epoch [90/200], Step [200/500] Loss:0.5137
Epoch [90/200], Step [300/500] Loss:0.1287
Epoch [90/200], Step [400/500] Loss:1.2059
Epoch [90/200], Step [500/500] Loss:1.1756
Accuracy of the model on the test images:92.48%
use 0.0min20.41212034225464s
Epoch [91/200], Step [100/500] Loss:1.2338
Epoch [91/200], Step [200/500] Loss:0.6848
Epoch [91/200], Step [300/500] Loss:1.0647
Epoch [91/200], Step [400/500] Loss:0.9616
Epoch [91/200], Step [500/500] Loss:0.9189
Accuracy of the model on the test images:93.32%
use 0.0min20.348811388015747s
Epoch [92/200], Step [100/500] Loss:1.1084
Epoch [92/200], Step [200/500] Loss:1.0042
Epoch [92/200], Step [300/500] Loss:0.4030
Epoch [92/200], Step [400/500] Loss:1.0330
Epoch [92/200], Step [500/500] Loss:0.9179
Accuracy of the model on the test images:92.81%
use 0.0min20.355735301971436s
Epoch [93/200], Step [100/500] Loss:0.7364
Epoch [93/200], Step [200/500] Loss:1.0574
Epoch [93/200], Step [300/500] Loss:0.6714
Epoch [93/200], Step [400/500] Loss:1.0800
Epoch [93/200], Step [500/500] Loss:0.7175
Accuracy of the model on the test images:92.89%
use 0.0min20.47357749938965s
Epoch [94/200], Step [100/500] Loss:0.5591
Epoch [94/200], Step [200/500] Loss:1.0056
Epoch [94/200], Step [300/500] Loss:1.0477
Epoch [94/200], Step [400/500] Loss:1.2228
Epoch [94/200], Step [500/500] Loss:0.8515
Accuracy of the model on the test images:92.65%
use 0.0min20.355154514312744s
Epoch [95/200], Step [100/500] Loss:0.8575
Epoch [95/200], Step [200/500] Loss:0.9975
Epoch [95/200], Step [300/500] Loss:1.1158
Epoch [95/200], Step [400/500] Loss:1.0921
Epoch [95/200], Step [500/500] Loss:0.9453
Accuracy of the model on the test images:93.45%
use 0.0min20.35475468635559s
Epoch [96/200], Step [100/500] Loss:0.5343
Epoch [96/200], Step [200/500] Loss:0.4544
Epoch [96/200], Step [300/500] Loss:0.7710
Epoch [96/200], Step [400/500] Loss:0.9027
Epoch [96/200], Step [500/500] Loss:0.7813
Accuracy of the model on the test images:93.24%
use 0.0min20.366315126419067s
Epoch [97/200], Step [100/500] Loss:1.0281
Epoch [97/200], Step [200/500] Loss:1.0061
Epoch [97/200], Step [300/500] Loss:1.0532
Epoch [97/200], Step [400/500] Loss:1.0409
Epoch [97/200], Step [500/500] Loss:1.0199
Accuracy of the model on the test images:92.55%
use 0.0min20.421039819717407s
Epoch [98/200], Step [100/500] Loss:0.8878
Epoch [98/200], Step [200/500] Loss:0.1803
Epoch [98/200], Step [300/500] Loss:0.7777
Epoch [98/200], Step [400/500] Loss:0.3656
Epoch [98/200], Step [500/500] Loss:1.1228
Accuracy of the model on the test images:93.74%
use 0.0min20.35983943939209s
Epoch [99/200], Step [100/500] Loss:1.1167
Epoch [99/200], Step [200/500] Loss:0.9838
Epoch [99/200], Step [300/500] Loss:0.6263
Epoch [99/200], Step [400/500] Loss:0.8882
Epoch [99/200], Step [500/500] Loss:1.0340
Accuracy of the model on the test images:92.63%
use 0.0min20.419955015182495s
Epoch [100/200], Step [100/500] Loss:0.7307
Epoch [100/200], Step [200/500] Loss:0.7769
Epoch [100/200], Step [300/500] Loss:0.2171
Epoch [100/200], Step [400/500] Loss:1.2046
Epoch [100/200], Step [500/500] Loss:1.0446
Accuracy of the model on the test images:93.14%
use 0.0min20.452274560928345s
Epoch [101/200], Step [100/500] Loss:0.8717
Epoch [101/200], Step [200/500] Loss:0.9948
Epoch [101/200], Step [300/500] Loss:1.1716
Epoch [101/200], Step [400/500] Loss:1.0076
Epoch [101/200], Step [500/500] Loss:0.7147
Accuracy of the model on the test images:93.89%
use 0.0min20.380756378173828s
Epoch [102/200], Step [100/500] Loss:1.1337
Epoch [102/200], Step [200/500] Loss:0.4186
Epoch [102/200], Step [300/500] Loss:0.9568
Epoch [102/200], Step [400/500] Loss:0.4825
Epoch [102/200], Step [500/500] Loss:0.3573
Accuracy of the model on the test images:93.62%
use 0.0min20.413810968399048s
Epoch [103/200], Step [100/500] Loss:1.0148
Epoch [103/200], Step [200/500] Loss:0.6628
Epoch [103/200], Step [300/500] Loss:1.0269
Epoch [103/200], Step [400/500] Loss:0.3809
Epoch [103/200], Step [500/500] Loss:0.7289
Accuracy of the model on the test images:95.05%
use 0.0min20.3670597076416s
Epoch [104/200], Step [100/500] Loss:0.9882
Epoch [104/200], Step [200/500] Loss:1.0050
Epoch [104/200], Step [300/500] Loss:0.9322
Epoch [104/200], Step [400/500] Loss:0.8870
Epoch [104/200], Step [500/500] Loss:0.2787
Accuracy of the model on the test images:95.08%
use 0.0min20.38065481185913s
Epoch [105/200], Step [100/500] Loss:0.4757
Epoch [105/200], Step [200/500] Loss:0.7992
Epoch [105/200], Step [300/500] Loss:0.7278
Epoch [105/200], Step [400/500] Loss:0.5998
Epoch [105/200], Step [500/500] Loss:0.7446
Accuracy of the model on the test images:94.98%
use 0.0min20.369545459747314s
Epoch [106/200], Step [100/500] Loss:0.7413
Epoch [106/200], Step [200/500] Loss:1.0611
Epoch [106/200], Step [300/500] Loss:0.9262
Epoch [106/200], Step [400/500] Loss:0.8101
Epoch [106/200], Step [500/500] Loss:0.7643
Accuracy of the model on the test images:95.22%
use 0.0min20.358132123947144s
Epoch [107/200], Step [100/500] Loss:0.7844
Epoch [107/200], Step [200/500] Loss:0.0925
Epoch [107/200], Step [300/500] Loss:0.5614
Epoch [107/200], Step [400/500] Loss:0.5424
Epoch [107/200], Step [500/500] Loss:0.7826
Accuracy of the model on the test images:95.27%
use 0.0min20.39941668510437s
Epoch [108/200], Step [100/500] Loss:0.2129
Epoch [108/200], Step [200/500] Loss:0.3992
Epoch [108/200], Step [300/500] Loss:0.8792
Epoch [108/200], Step [400/500] Loss:0.7090
Epoch [108/200], Step [500/500] Loss:0.6356
Accuracy of the model on the test images:95.29%
use 0.0min20.351104259490967s
Epoch [109/200], Step [100/500] Loss:0.6356
Epoch [109/200], Step [200/500] Loss:0.8048
Epoch [109/200], Step [300/500] Loss:0.1197
Epoch [109/200], Step [400/500] Loss:0.5651
Epoch [109/200], Step [500/500] Loss:0.8223
Accuracy of the model on the test images:95.14%
use 0.0min20.34543776512146s
Epoch [110/200], Step [100/500] Loss:0.6938
Epoch [110/200], Step [200/500] Loss:0.4959
Epoch [110/200], Step [300/500] Loss:0.8740
Epoch [110/200], Step [400/500] Loss:0.7046
Epoch [110/200], Step [500/500] Loss:0.7780
Accuracy of the model on the test images:95.06%
use 0.0min20.357868432998657s
Epoch [111/200], Step [100/500] Loss:0.8586
Epoch [111/200], Step [200/500] Loss:0.9439
Epoch [111/200], Step [300/500] Loss:0.9383
Epoch [111/200], Step [400/500] Loss:0.7799
Epoch [111/200], Step [500/500] Loss:0.7568
Accuracy of the model on the test images:95.04%
use 0.0min20.377333164215088s
Epoch [112/200], Step [100/500] Loss:0.8705
Epoch [112/200], Step [200/500] Loss:0.4833
Epoch [112/200], Step [300/500] Loss:0.7744
Epoch [112/200], Step [400/500] Loss:0.8568
Epoch [112/200], Step [500/500] Loss:0.6183
Accuracy of the model on the test images:95.12%
use 0.0min20.358330488204956s
Epoch [113/200], Step [100/500] Loss:0.8118
Epoch [113/200], Step [200/500] Loss:0.2100
Epoch [113/200], Step [300/500] Loss:0.9770
Epoch [113/200], Step [400/500] Loss:0.2692
Epoch [113/200], Step [500/500] Loss:1.0045
Accuracy of the model on the test images:95.13%
use 0.0min20.3507342338562s
Epoch [114/200], Step [100/500] Loss:0.7069
Epoch [114/200], Step [200/500] Loss:0.8443
Epoch [114/200], Step [300/500] Loss:0.8858
Epoch [114/200], Step [400/500] Loss:0.8029
Epoch [114/200], Step [500/500] Loss:0.9669
Accuracy of the model on the test images:95.19%
use 0.0min20.42878007888794s
Epoch [115/200], Step [100/500] Loss:0.5202
Epoch [115/200], Step [200/500] Loss:0.0889
Epoch [115/200], Step [300/500] Loss:0.5234
Epoch [115/200], Step [400/500] Loss:0.2499
Epoch [115/200], Step [500/500] Loss:0.2063
Accuracy of the model on the test images:95.3%