pytorch-study/14.ipynb

In [1]:

import numpy as np
import random
import matplotlib.pyplot as plt
from torch.utils.data import DataLoader

In [3]:

w = 2
b = 3
xlim = [-10, 10]
x_train = np.random.randint(low=xlim[0], high=xlim[1], size=30)

y_train = [w * x + b + random.randint(0, 2) for x in x_train]

plt.plot(x_train, y_train, 'bo')

Out[3]:

[<matplotlib.lines.Line2D at 0x7b28d4b99330>]

In [4]:

import torch
import torch.nn as nn


class LinearModel(nn.Module):
    def __init__(self):
        super().__init__()
        self.weight = nn.Parameter(torch.randn(1))
        self.bias = nn.Parameter(torch.randn(1))

    def forward(self, input):
        return (input * self.weight) + self.bias

In [6]:

model = LinearModel()
x = torch.tensor(3)
y = model(x)

In [9]:

model = LinearModel()

# 定义优化器
optimizer = torch.optim.SGD(model.parameters(), lr=1e-4, weight_decay=1e-2, momentum=0.9)

y_train = torch.tensor(y_train, dtype=torch.float32)

# 训练模型
for epoch in range(1000):
    input = torch.from_numpy(x_train)
    output = model(input)
    loss = nn.functional.mse_loss(output, y_train)
    optimizer.zero_grad()
    loss.backward()
    optimizer.step()

/tmp/ipykernel_9/825497026.py:6: UserWarning: To copy construct from a tensor, it is recommended to use sourceTensor.clone().detach() or sourceTensor.clone().detach().requires_grad_(True), rather than torch.tensor(sourceTensor).
  y_train = torch.tensor(y_train, dtype=torch.float32)

In [11]:

for param in model.parameters():
    print(param)

Parameter containing:
tensor([2.0042], requires_grad=True)
Parameter containing:
tensor([3.4570], requires_grad=True)

In [12]:

class CustomLayer(nn.Module):
    def __init__(self, in_features, out_features):
        super().__init__()
        self.weight = nn.Parameter(torch.randn(in_features, out_features))
        self.bias = nn.Parameter(torch.randn(out_features))

    def forward(self, input):
        return input @ self.weight + self.bias

In [15]:

print(model.state_dict())
torch.save(model.state_dict(), './data/linear_model.pth')

OrderedDict([('weight', tensor([2.0042])), ('bias', tensor([3.4570]))])

In [16]:

# 定义网络结构
linear_model = LinearModel()
# 加载保存的参数
linear_model.load_state_dict(torch.load('./data/linear_model.pth'))
linear_model.eval()
for param in linear_model.parameters():
    print(param)

Parameter containing:
tensor([2.0042], requires_grad=True)
Parameter containing:
tensor([3.4570], requires_grad=True)

In [22]:

import os

os.environ['TORCH_HOME'] = './data'  # 修改存储目录

import torch

print(torch.hub.get_dir())  # 默认是 ~/.cache/torch

./data/hub

In [43]:

import torchvision.models as models

alexnet = models.alexnet(pretrained=True)

In [47]:

from PIL import Image
import torchvision
from torchvision import transforms

im = Image.open('./data/images/dog.jpg')

transform = transforms.Compose([
    transforms.Resize((224, 224)),
    transforms.ToTensor(),
    transforms.Normalize(mean=[0.485, 0.456, 0.406], std=[0.229, 0.224, 0.225])
])

input_tensor = transform(im).unsqueeze(0)
alexnet.eval()
alexnet(input_tensor).argmax()

Out[47]:

tensor(263)

In [27]:

cifar10_dataset = torchvision.datasets.CIFAR10(root='./data', train=False, transform=transforms.ToTensor(),
                                               target_transform=None, download=True)

Using downloaded and verified file: ./data/cifar-10-python.tar.gz
Extracting ./data/cifar-10-python.tar.gz to ./data

In [30]:

from torch.utils.data import DataLoader

tensor_dataloader = DataLoader(cifar10_dataset, batch_size=48)

data_iter = iter(tensor_dataloader)
images, labels = next(data_iter)
print(images.shape)

grid_tensor = torchvision.utils.make_grid(images, nrow=16, padding=2)
grid_img = transforms.ToPILImage()(grid_tensor)
grid_img.show()

torch.Size([48, 3, 32, 32])

In [31]:

print(alexnet)

AlexNet(
  (features): Sequential(
    (0): Conv2d(3, 64, kernel_size=(11, 11), stride=(4, 4), padding=(2, 2))
    (1): ReLU(inplace=True)
    (2): MaxPool2d(kernel_size=3, stride=2, padding=0, dilation=1, ceil_mode=False)
    (3): Conv2d(64, 192, kernel_size=(5, 5), stride=(1, 1), padding=(2, 2))
    (4): ReLU(inplace=True)
    (5): MaxPool2d(kernel_size=3, stride=2, padding=0, dilation=1, ceil_mode=False)
    (6): Conv2d(192, 384, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))
    (7): ReLU(inplace=True)
    (8): Conv2d(384, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))
    (9): ReLU(inplace=True)
    (10): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))
    (11): ReLU(inplace=True)
    (12): MaxPool2d(kernel_size=3, stride=2, padding=0, dilation=1, ceil_mode=False)
  )
  (avgpool): AdaptiveAvgPool2d(output_size=(6, 6))
  (classifier): Sequential(
    (0): Dropout(p=0.5, inplace=False)
    (1): Linear(in_features=9216, out_features=4096, bias=True)
    (2): ReLU(inplace=True)
    (3): Dropout(p=0.5, inplace=False)
    (4): Linear(in_features=4096, out_features=4096, bias=True)
    (5): ReLU(inplace=True)
    (6): Linear(in_features=4096, out_features=1000, bias=True)
  )
)

In [32]:

# 提取分类层的输入参数
fc_in_features = alexnet.classifier[6].in_features

# 修改与训练模型的输出分类数
alexnet.classifier[6] = nn.Linear(fc_in_features, 10)

print(alexnet)

AlexNet(
  (features): Sequential(
    (0): Conv2d(3, 64, kernel_size=(11, 11), stride=(4, 4), padding=(2, 2))
    (1): ReLU(inplace=True)
    (2): MaxPool2d(kernel_size=3, stride=2, padding=0, dilation=1, ceil_mode=False)
    (3): Conv2d(64, 192, kernel_size=(5, 5), stride=(1, 1), padding=(2, 2))
    (4): ReLU(inplace=True)
    (5): MaxPool2d(kernel_size=3, stride=2, padding=0, dilation=1, ceil_mode=False)
    (6): Conv2d(192, 384, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))
    (7): ReLU(inplace=True)
    (8): Conv2d(384, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))
    (9): ReLU(inplace=True)
    (10): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))
    (11): ReLU(inplace=True)
    (12): MaxPool2d(kernel_size=3, stride=2, padding=0, dilation=1, ceil_mode=False)
  )
  (avgpool): AdaptiveAvgPool2d(output_size=(6, 6))
  (classifier): Sequential(
    (0): Dropout(p=0.5, inplace=False)
    (1): Linear(in_features=9216, out_features=4096, bias=True)
    (2): ReLU(inplace=True)
    (3): Dropout(p=0.5, inplace=False)
    (4): Linear(in_features=4096, out_features=4096, bias=True)
    (5): ReLU(inplace=True)
    (6): Linear(in_features=4096, out_features=10, bias=True)
  )
)

In [45]:

transform = transforms.Compose([
    transforms.RandomResizedCrop((224, 224)),
    transforms.ToTensor(),
    transforms.Normalize(mean=[0.485, 0.456, 0.406], std=[0.229, 0.224, 0.225])
])
cifar10_dataset = torchvision.datasets.CIFAR10(root='./data', train=False, transform=transform,
                                               target_transform=None, download=True)

dataloader = DataLoader(cifar10_dataset, batch_size=32, shuffle=True, num_workers=0)

Files already downloaded and verified

In [46]:

optimizer = torch.optim.SGD(alexnet.parameters(), lr=1e-4, weight_decay=1e-2, momentum=0.9)

In [48]:

# 训练3个Epoch
for epoch in range(3):
    for item in dataloader:
        output = alexnet(item[0])
        target = item[1]
        # 使用交叉熵损失函数
        loss = nn.CrossEntropyLoss()(output, target)
        print('Epoch {}, Loss {}'.format(epoch + 1, loss))
        #以下代码的含义，我们在之前的文章中已经介绍过了
        alexnet.zero_grad()
        loss.backward()
        optimizer.step()

Epoch 1, Loss 11.390161514282227
Epoch 1, Loss 11.10473346710205
Epoch 1, Loss 9.650487899780273
Epoch 1, Loss 8.740126609802246
Epoch 1, Loss 8.255596160888672
Epoch 1, Loss 8.651772499084473
Epoch 1, Loss 7.846767425537109
Epoch 1, Loss 7.975433826446533
Epoch 1, Loss 7.931630611419678
Epoch 1, Loss 7.305787563323975
Epoch 1, Loss 7.42929744720459
Epoch 1, Loss 7.351666450500488
Epoch 1, Loss 7.194794178009033
Epoch 1, Loss 6.7262864112854
Epoch 1, Loss 7.113491535186768
Epoch 1, Loss 6.258823394775391
Epoch 1, Loss 6.161075115203857
Epoch 1, Loss 6.2522382736206055
Epoch 1, Loss 5.654983997344971
Epoch 1, Loss 5.217900276184082
Epoch 1, Loss 4.8547234535217285
Epoch 1, Loss 5.045372486114502
Epoch 1, Loss 5.413811206817627
Epoch 1, Loss 5.163278579711914
Epoch 1, Loss 5.05932092666626
Epoch 1, Loss 4.755397796630859
Epoch 1, Loss 5.098104476928711
Epoch 1, Loss 4.68168830871582
Epoch 1, Loss 4.83888578414917
Epoch 1, Loss 4.25895357131958
Epoch 1, Loss 4.044596195220947
Epoch 1, Loss 4.3518500328063965
Epoch 1, Loss 4.016391277313232
Epoch 1, Loss 3.4523468017578125
Epoch 1, Loss 3.6078708171844482
Epoch 1, Loss 3.3324170112609863
Epoch 1, Loss 3.146799087524414
Epoch 1, Loss 3.3123574256896973
Epoch 1, Loss 2.9692232608795166
Epoch 1, Loss 3.150871992111206
Epoch 1, Loss 3.1080782413482666
Epoch 1, Loss 2.697796583175659
Epoch 1, Loss 2.8308873176574707
Epoch 1, Loss 2.4798989295959473
Epoch 1, Loss 3.044487237930298
Epoch 1, Loss 2.5619263648986816
Epoch 1, Loss 3.03995943069458
Epoch 1, Loss 2.5703911781311035
Epoch 1, Loss 2.622386932373047
Epoch 1, Loss 2.5821354389190674
Epoch 1, Loss 2.2508366107940674
Epoch 1, Loss 2.5529377460479736
Epoch 1, Loss 2.8578474521636963
Epoch 1, Loss 2.4568989276885986
Epoch 1, Loss 2.5013320446014404
Epoch 1, Loss 1.9693650007247925
Epoch 1, Loss 2.0936081409454346
Epoch 1, Loss 2.420494794845581
Epoch 1, Loss 2.0740833282470703
Epoch 1, Loss 2.5092103481292725
Epoch 1, Loss 2.2415835857391357
Epoch 1, Loss 2.1151254177093506
Epoch 1, Loss 2.4123313426971436
Epoch 1, Loss 2.344158172607422
Epoch 1, Loss 2.31493878364563
Epoch 1, Loss 2.190805673599243
Epoch 1, Loss 2.318197250366211
Epoch 1, Loss 2.216266393661499
Epoch 1, Loss 2.1666195392608643
Epoch 1, Loss 2.1229658126831055
Epoch 1, Loss 2.004812717437744
Epoch 1, Loss 2.013103723526001
Epoch 1, Loss 2.356696844100952
Epoch 1, Loss 2.244671106338501
Epoch 1, Loss 2.1417644023895264
Epoch 1, Loss 2.158512592315674
Epoch 1, Loss 2.3859620094299316
Epoch 1, Loss 2.2160897254943848
Epoch 1, Loss 2.147921562194824
Epoch 1, Loss 1.97897469997406
Epoch 1, Loss 1.9645150899887085
Epoch 1, Loss 2.23921799659729
Epoch 1, Loss 1.6986244916915894
Epoch 1, Loss 2.0967540740966797
Epoch 1, Loss 1.9966078996658325
Epoch 1, Loss 2.2282333374023438
Epoch 1, Loss 2.05538010597229
Epoch 1, Loss 2.0945675373077393
Epoch 1, Loss 2.4314029216766357
Epoch 1, Loss 2.187045097351074
Epoch 1, Loss 2.196854591369629
Epoch 1, Loss 1.9174386262893677
Epoch 1, Loss 1.839700698852539
Epoch 1, Loss 2.2496514320373535
Epoch 1, Loss 1.8240433931350708
Epoch 1, Loss 1.7558871507644653
Epoch 1, Loss 2.0411672592163086
Epoch 1, Loss 2.1743698120117188
Epoch 1, Loss 2.2457425594329834
Epoch 1, Loss 2.170886278152466
Epoch 1, Loss 2.0813663005828857
Epoch 1, Loss 2.0077240467071533
Epoch 1, Loss 2.152672290802002
Epoch 1, Loss 1.8744524717330933
Epoch 1, Loss 1.8593156337738037
Epoch 1, Loss 2.0026607513427734
Epoch 1, Loss 1.997119665145874
Epoch 1, Loss 1.9461400508880615
Epoch 1, Loss 1.8490214347839355
Epoch 1, Loss 1.8053253889083862
Epoch 1, Loss 1.8681100606918335
Epoch 1, Loss 1.7914729118347168
Epoch 1, Loss 1.9788165092468262
Epoch 1, Loss 2.0497021675109863
Epoch 1, Loss 1.9876694679260254
Epoch 1, Loss 1.8120652437210083
Epoch 1, Loss 1.7909998893737793
Epoch 1, Loss 1.8859608173370361
Epoch 1, Loss 1.8429142236709595
Epoch 1, Loss 2.0022695064544678
Epoch 1, Loss 1.733254313468933
Epoch 1, Loss 1.7776328325271606
Epoch 1, Loss 1.9475072622299194
Epoch 1, Loss 1.829969048500061
Epoch 1, Loss 1.915332555770874
Epoch 1, Loss 2.0018343925476074
Epoch 1, Loss 1.909234642982483
Epoch 1, Loss 1.781591773033142
Epoch 1, Loss 1.859167456626892
Epoch 1, Loss 2.129415988922119
Epoch 1, Loss 1.7759991884231567
Epoch 1, Loss 1.9012956619262695
Epoch 1, Loss 1.93148934841156
Epoch 1, Loss 1.9490193128585815
Epoch 1, Loss 1.9708281755447388
Epoch 1, Loss 1.9900016784667969
Epoch 1, Loss 2.2335097789764404
Epoch 1, Loss 2.0086076259613037
Epoch 1, Loss 1.91185462474823
Epoch 1, Loss 2.0486092567443848
Epoch 1, Loss 1.8756918907165527
Epoch 1, Loss 1.922384262084961
Epoch 1, Loss 1.9162771701812744
Epoch 1, Loss 2.30863881111145
Epoch 1, Loss 1.7405242919921875
Epoch 1, Loss 2.072435140609741
Epoch 1, Loss 1.9467562437057495
Epoch 1, Loss 1.6027220487594604
Epoch 1, Loss 1.8665951490402222
Epoch 1, Loss 1.8461048603057861
Epoch 1, Loss 1.788988709449768
Epoch 1, Loss 1.9649192094802856
Epoch 1, Loss 1.8493802547454834
Epoch 1, Loss 2.090841293334961
Epoch 1, Loss 2.06889009475708
Epoch 1, Loss 1.6090549230575562
Epoch 1, Loss 1.6090567111968994
Epoch 1, Loss 2.033944606781006
Epoch 1, Loss 1.574442744255066
Epoch 1, Loss 1.8012334108352661
Epoch 1, Loss 1.911574363708496
Epoch 1, Loss 1.6573156118392944
Epoch 1, Loss 1.808171033859253
Epoch 1, Loss 2.030792474746704
Epoch 1, Loss 1.7114735841751099
Epoch 1, Loss 1.6729048490524292
Epoch 1, Loss 1.8977779150009155
Epoch 1, Loss 2.01203989982605
Epoch 1, Loss 1.9247945547103882
Epoch 1, Loss 1.9593578577041626
Epoch 1, Loss 2.00467586517334
Epoch 1, Loss 1.7006726264953613
Epoch 1, Loss 1.7836061716079712
Epoch 1, Loss 1.7102396488189697
Epoch 1, Loss 1.7951672077178955
Epoch 1, Loss 1.9334559440612793
Epoch 1, Loss 1.8003500699996948
Epoch 1, Loss 1.7520798444747925
Epoch 1, Loss 1.7685191631317139
Epoch 1, Loss 1.9277746677398682
Epoch 1, Loss 2.1595067977905273
Epoch 1, Loss 1.7749875783920288
Epoch 1, Loss 1.8476790189743042
Epoch 1, Loss 1.740425705909729
Epoch 1, Loss 1.7416574954986572
Epoch 1, Loss 1.7841198444366455
Epoch 1, Loss 1.9414969682693481
Epoch 1, Loss 1.6255710124969482
Epoch 1, Loss 1.76301908493042
Epoch 1, Loss 1.7434982061386108
Epoch 1, Loss 1.4769729375839233
Epoch 1, Loss 1.6410868167877197
Epoch 1, Loss 1.984968900680542
Epoch 1, Loss 1.8557807207107544
Epoch 1, Loss 1.8334715366363525
Epoch 1, Loss 1.6681809425354004
Epoch 1, Loss 1.9186973571777344
Epoch 1, Loss 1.6244840621948242
Epoch 1, Loss 1.8782172203063965
Epoch 1, Loss 1.9548263549804688
Epoch 1, Loss 1.808607816696167
Epoch 1, Loss 1.7929818630218506
Epoch 1, Loss 1.7395025491714478
Epoch 1, Loss 1.6287509202957153
Epoch 1, Loss 1.854030728340149
Epoch 1, Loss 1.7849431037902832
Epoch 1, Loss 1.6917438507080078
Epoch 1, Loss 1.7881979942321777
Epoch 1, Loss 1.8737754821777344
Epoch 1, Loss 1.9840139150619507
Epoch 1, Loss 2.040748119354248
Epoch 1, Loss 1.784123182296753
Epoch 1, Loss 1.8795673847198486
Epoch 1, Loss 1.909524917602539
Epoch 1, Loss 1.7231237888336182
Epoch 1, Loss 1.7076764106750488
Epoch 1, Loss 1.5988054275512695
Epoch 1, Loss 1.7095640897750854
Epoch 1, Loss 1.8038264513015747
Epoch 1, Loss 1.3908846378326416
Epoch 1, Loss 1.8790314197540283
Epoch 1, Loss 1.624179482460022
Epoch 1, Loss 1.5522247552871704
Epoch 1, Loss 2.0716910362243652
Epoch 1, Loss 1.758894443511963
Epoch 1, Loss 1.4196009635925293
Epoch 1, Loss 1.754120111465454
Epoch 1, Loss 1.631662368774414
Epoch 1, Loss 2.017249345779419
Epoch 1, Loss 1.791286826133728
Epoch 1, Loss 1.648071050643921
Epoch 1, Loss 2.0191264152526855
Epoch 1, Loss 1.5867856740951538
Epoch 1, Loss 1.5980314016342163
Epoch 1, Loss 1.792881965637207
Epoch 1, Loss 1.6422427892684937
Epoch 1, Loss 1.5760972499847412
Epoch 1, Loss 1.7619928121566772
Epoch 1, Loss 1.7108908891677856
Epoch 1, Loss 1.5352228879928589
Epoch 1, Loss 1.8060669898986816
Epoch 1, Loss 1.6254026889801025
Epoch 1, Loss 1.5435885190963745
Epoch 1, Loss 1.754928708076477
Epoch 1, Loss 1.6306538581848145
Epoch 1, Loss 1.4022413492202759
Epoch 1, Loss 1.9609127044677734
Epoch 1, Loss 1.795851469039917
Epoch 1, Loss 2.1257569789886475
Epoch 1, Loss 1.8683059215545654
Epoch 1, Loss 1.6383010149002075
Epoch 1, Loss 1.6648447513580322
Epoch 1, Loss 1.6740763187408447
Epoch 1, Loss 1.7074679136276245
Epoch 1, Loss 1.63435697555542
Epoch 1, Loss 1.6114144325256348
Epoch 1, Loss 1.9356369972229004
Epoch 1, Loss 1.4554531574249268
Epoch 1, Loss 1.6876460313796997
Epoch 1, Loss 1.844913363456726
Epoch 1, Loss 1.6566756963729858
Epoch 1, Loss 1.7295408248901367
Epoch 1, Loss 1.9728142023086548
Epoch 1, Loss 1.5791298151016235
Epoch 1, Loss 1.505454659461975
Epoch 1, Loss 1.6260864734649658
Epoch 1, Loss 1.7501051425933838
Epoch 1, Loss 1.8571727275848389
Epoch 1, Loss 1.8610074520111084
Epoch 1, Loss 1.7246692180633545
Epoch 1, Loss 1.8025296926498413
Epoch 1, Loss 1.6552321910858154
Epoch 1, Loss 1.803894281387329
Epoch 1, Loss 1.7384475469589233
Epoch 1, Loss 1.6529415845870972
Epoch 1, Loss 1.723402976989746
Epoch 1, Loss 1.6311922073364258
Epoch 1, Loss 1.645974040031433
Epoch 1, Loss 2.0301053524017334
Epoch 1, Loss 1.880918264389038
Epoch 1, Loss 1.6266663074493408
Epoch 1, Loss 1.5199600458145142
Epoch 1, Loss 1.4093431234359741
Epoch 1, Loss 1.8859593868255615
Epoch 1, Loss 1.752493143081665
Epoch 1, Loss 1.6518990993499756
Epoch 1, Loss 1.631445050239563
Epoch 1, Loss 1.512643575668335
Epoch 1, Loss 1.6602154970169067
Epoch 1, Loss 1.5975916385650635
Epoch 1, Loss 1.6646537780761719
Epoch 1, Loss 1.8574987649917603
Epoch 1, Loss 1.8242039680480957
Epoch 1, Loss 1.674977421760559
Epoch 1, Loss 1.8695521354675293
Epoch 1, Loss 1.6858590841293335
Epoch 1, Loss 1.7875169515609741
Epoch 1, Loss 1.5225797891616821
Epoch 1, Loss 1.5076755285263062
Epoch 1, Loss 1.5932300090789795
Epoch 1, Loss 1.5611786842346191
Epoch 1, Loss 1.8602344989776611
Epoch 1, Loss 1.609197974205017
Epoch 1, Loss 1.4913467168807983
Epoch 1, Loss 1.8280011415481567
Epoch 1, Loss 1.3795815706253052
Epoch 1, Loss 1.8851159811019897
Epoch 1, Loss 1.7245428562164307
Epoch 1, Loss 1.7438805103302002
Epoch 1, Loss 1.691540002822876
Epoch 1, Loss 1.842980980873108
Epoch 1, Loss 2.019976854324341
Epoch 1, Loss 1.6800366640090942
Epoch 2, Loss 1.300485372543335
Epoch 2, Loss 1.784948706626892
Epoch 2, Loss 1.4291561841964722
Epoch 2, Loss 2.0613596439361572
Epoch 2, Loss 1.7472537755966187
Epoch 2, Loss 1.8874013423919678
Epoch 2, Loss 1.653011679649353
Epoch 2, Loss 1.616155743598938
Epoch 2, Loss 1.6922290325164795
Epoch 2, Loss 2.1530885696411133
Epoch 2, Loss 2.0140461921691895
Epoch 2, Loss 1.6550230979919434
Epoch 2, Loss 1.909831166267395
Epoch 2, Loss 1.6249158382415771
Epoch 2, Loss 1.6772966384887695
Epoch 2, Loss 1.4758840799331665
Epoch 2, Loss 1.3574433326721191
Epoch 2, Loss 1.6426734924316406
Epoch 2, Loss 1.656657099723816
Epoch 2, Loss 1.6859227418899536
Epoch 2, Loss 1.7225008010864258
Epoch 2, Loss 1.7683744430541992
Epoch 2, Loss 1.50262451171875
Epoch 2, Loss 1.4230449199676514
Epoch 2, Loss 2.0901670455932617
Epoch 2, Loss 1.8206089735031128
Epoch 2, Loss 1.5206626653671265
Epoch 2, Loss 1.2650506496429443
Epoch 2, Loss 1.6379344463348389
Epoch 2, Loss 1.4596773386001587
Epoch 2, Loss 1.6611775159835815
Epoch 2, Loss 1.4224178791046143
Epoch 2, Loss 1.579258918762207
Epoch 2, Loss 1.6873137950897217
Epoch 2, Loss 1.6964925527572632
Epoch 2, Loss 1.609260082244873
Epoch 2, Loss 1.7320889234542847
Epoch 2, Loss 1.6290900707244873
Epoch 2, Loss 1.293398141860962
Epoch 2, Loss 1.4279762506484985
Epoch 2, Loss 1.3718245029449463
Epoch 2, Loss 1.5147663354873657
Epoch 2, Loss 1.6692233085632324
Epoch 2, Loss 1.9885149002075195
Epoch 2, Loss 1.2877438068389893
Epoch 2, Loss 1.4702204465866089
Epoch 2, Loss 1.2480032444000244
Epoch 2, Loss 1.434686541557312
Epoch 2, Loss 1.6788338422775269
Epoch 2, Loss 1.6272554397583008
Epoch 2, Loss 1.740920901298523
Epoch 2, Loss 1.6473010778427124
Epoch 2, Loss 1.8052456378936768
Epoch 2, Loss 1.6307520866394043
Epoch 2, Loss 1.8539469242095947
Epoch 2, Loss 1.6478047370910645
Epoch 2, Loss 1.4329921007156372
Epoch 2, Loss 1.539477825164795
Epoch 2, Loss 1.424522042274475
Epoch 2, Loss 1.4489614963531494
Epoch 2, Loss 1.7238457202911377
Epoch 2, Loss 1.6953550577163696
Epoch 2, Loss 1.5163968801498413
Epoch 2, Loss 1.5332549810409546
Epoch 2, Loss 1.6442549228668213
Epoch 2, Loss 1.4431391954421997
Epoch 2, Loss 1.6149733066558838
Epoch 2, Loss 1.5853711366653442
Epoch 2, Loss 1.870877981185913
Epoch 2, Loss 1.1911654472351074
Epoch 2, Loss 1.7067879438400269
Epoch 2, Loss 1.882607340812683
Epoch 2, Loss 1.818520188331604
Epoch 2, Loss 1.7367340326309204
Epoch 2, Loss 1.5798382759094238
Epoch 2, Loss 1.4751096963882446
Epoch 2, Loss 1.6802865266799927
Epoch 2, Loss 1.6236398220062256
Epoch 2, Loss 1.636420488357544
Epoch 2, Loss 1.4924060106277466
Epoch 2, Loss 1.691157341003418
Epoch 2, Loss 1.5413718223571777
Epoch 2, Loss 2.1446926593780518
Epoch 2, Loss 1.8346779346466064
Epoch 2, Loss 2.103973150253296
Epoch 2, Loss 1.7558985948562622
Epoch 2, Loss 1.6326268911361694
Epoch 2, Loss 1.768365502357483
Epoch 2, Loss 1.7573071718215942
Epoch 2, Loss 1.3856499195098877
Epoch 2, Loss 1.5004853010177612
Epoch 2, Loss 1.6939235925674438
Epoch 2, Loss 1.6444960832595825
Epoch 2, Loss 1.6750274896621704
Epoch 2, Loss 1.919832468032837
Epoch 2, Loss 1.9656294584274292
Epoch 2, Loss 1.7102035284042358
Epoch 2, Loss 1.6626485586166382
Epoch 2, Loss 1.5556167364120483
Epoch 2, Loss 1.5860583782196045
Epoch 2, Loss 1.595888614654541
Epoch 2, Loss 1.5543204545974731
Epoch 2, Loss 1.6619356870651245
Epoch 2, Loss 1.6742863655090332
Epoch 2, Loss 1.6100677251815796
Epoch 2, Loss 1.7293599843978882
Epoch 2, Loss 1.758224368095398
Epoch 2, Loss 1.6590352058410645
Epoch 2, Loss 1.6913721561431885
Epoch 2, Loss 1.4077750444412231
Epoch 2, Loss 1.1626904010772705
Epoch 2, Loss 1.7251015901565552
Epoch 2, Loss 1.2510126829147339
Epoch 2, Loss 1.9657052755355835
Epoch 2, Loss 1.4909043312072754
Epoch 2, Loss 1.4821815490722656
Epoch 2, Loss 1.382213830947876
Epoch 2, Loss 1.548332691192627
Epoch 2, Loss 1.6348422765731812
Epoch 2, Loss 1.2901160717010498
Epoch 2, Loss 1.7978826761245728
Epoch 2, Loss 1.6333582401275635
Epoch 2, Loss 1.7142982482910156
Epoch 2, Loss 1.3097147941589355
Epoch 2, Loss 1.648378610610962
Epoch 2, Loss 1.5350890159606934
Epoch 2, Loss 1.2843588590621948
Epoch 2, Loss 1.7580673694610596
Epoch 2, Loss 1.307361125946045
Epoch 2, Loss 1.6136400699615479
Epoch 2, Loss 1.7835789918899536
Epoch 2, Loss 1.8142532110214233
Epoch 2, Loss 1.58303701877594
Epoch 2, Loss 1.6874006986618042
Epoch 2, Loss 1.3635849952697754
Epoch 2, Loss 1.7432588338851929
Epoch 2, Loss 1.512610673904419
Epoch 2, Loss 1.7223109006881714
Epoch 2, Loss 1.7032732963562012
Epoch 2, Loss 1.478933334350586
Epoch 2, Loss 1.613121747970581
Epoch 2, Loss 2.0838115215301514
Epoch 2, Loss 1.5958144664764404
Epoch 2, Loss 1.2433825731277466
Epoch 2, Loss 1.8604488372802734
Epoch 2, Loss 1.7540366649627686
Epoch 2, Loss 1.6867806911468506
Epoch 2, Loss 1.2986611127853394
Epoch 2, Loss 1.0995105504989624
Epoch 2, Loss 1.5499902963638306
Epoch 2, Loss 1.5567690134048462
Epoch 2, Loss 1.38352632522583
Epoch 2, Loss 1.5722194910049438
Epoch 2, Loss 1.2573655843734741
Epoch 2, Loss 1.7265040874481201
Epoch 2, Loss 1.432693362236023
Epoch 2, Loss 1.768494963645935
Epoch 2, Loss 1.4405817985534668
Epoch 2, Loss 1.6624199151992798
Epoch 2, Loss 1.4682323932647705
Epoch 2, Loss 1.5353881120681763
Epoch 2, Loss 1.3602796792984009
Epoch 2, Loss 1.5736920833587646
Epoch 2, Loss 2.03515625
Epoch 2, Loss 1.3737272024154663
Epoch 2, Loss 1.8202908039093018
Epoch 2, Loss 2.0836384296417236
Epoch 2, Loss 1.6040118932724
Epoch 2, Loss 1.3748340606689453
Epoch 2, Loss 1.4863075017929077
Epoch 2, Loss 1.499975562095642
Epoch 2, Loss 1.3765817880630493
Epoch 2, Loss 1.457093596458435
Epoch 2, Loss 1.6555736064910889
Epoch 2, Loss 1.4382950067520142
Epoch 2, Loss 1.453316569328308
Epoch 2, Loss 1.5730074644088745
Epoch 2, Loss 1.7737727165222168
Epoch 2, Loss 1.6523839235305786
Epoch 2, Loss 1.3266234397888184
Epoch 2, Loss 1.6159758567810059
Epoch 2, Loss 1.4704073667526245
Epoch 2, Loss 1.531602144241333
Epoch 2, Loss 2.0312156677246094
Epoch 2, Loss 1.888283133506775
Epoch 2, Loss 1.3371005058288574
Epoch 2, Loss 1.3945879936218262
Epoch 2, Loss 1.529701590538025
Epoch 2, Loss 1.7100629806518555
Epoch 2, Loss 1.3226559162139893
Epoch 2, Loss 1.8098974227905273
Epoch 2, Loss 1.7791922092437744
Epoch 2, Loss 1.6193798780441284
Epoch 2, Loss 1.859694242477417
Epoch 2, Loss 1.5189138650894165
Epoch 2, Loss 1.415958046913147
Epoch 2, Loss 1.212141752243042
Epoch 2, Loss 1.6975561380386353
Epoch 2, Loss 1.7593353986740112
Epoch 2, Loss 1.1995075941085815
Epoch 2, Loss 1.359512448310852
Epoch 2, Loss 1.4644752740859985
Epoch 2, Loss 1.4567301273345947
Epoch 2, Loss 1.4640811681747437
Epoch 2, Loss 1.5118407011032104
Epoch 2, Loss 1.3671762943267822
Epoch 2, Loss 1.4093780517578125
Epoch 2, Loss 1.6929950714111328
Epoch 2, Loss 1.421968698501587
Epoch 2, Loss 1.3280011415481567
Epoch 2, Loss 1.5840849876403809
Epoch 2, Loss 1.3747957944869995
Epoch 2, Loss 1.6173515319824219
Epoch 2, Loss 1.3797783851623535
Epoch 2, Loss 1.245814323425293
Epoch 2, Loss 1.311650276184082
Epoch 2, Loss 1.570831298828125
Epoch 2, Loss 1.7656368017196655
Epoch 2, Loss 1.4017724990844727
Epoch 2, Loss 1.1819841861724854
Epoch 2, Loss 1.2329351902008057
Epoch 2, Loss 1.3429627418518066
Epoch 2, Loss 1.5433602333068848
Epoch 2, Loss 1.2263259887695312
Epoch 2, Loss 1.44208562374115
Epoch 2, Loss 1.5704418420791626
Epoch 2, Loss 1.8433754444122314
Epoch 2, Loss 1.3556145429611206
Epoch 2, Loss 1.8337855339050293
Epoch 2, Loss 1.3890026807785034
Epoch 2, Loss 1.2915153503417969
Epoch 2, Loss 1.2885075807571411
Epoch 2, Loss 1.626181960105896
Epoch 2, Loss 1.3702009916305542
Epoch 2, Loss 1.5307343006134033
Epoch 2, Loss 1.5089809894561768
Epoch 2, Loss 1.5458707809448242
Epoch 2, Loss 1.6739248037338257
Epoch 2, Loss 1.6757007837295532
Epoch 2, Loss 1.6528023481369019
Epoch 2, Loss 2.0662283897399902
Epoch 2, Loss 1.6763783693313599
Epoch 2, Loss 1.511576771736145
Epoch 2, Loss 1.5105756521224976
Epoch 2, Loss 1.6672422885894775
Epoch 2, Loss 1.3297759294509888
Epoch 2, Loss 1.3444056510925293
Epoch 2, Loss 1.8276199102401733
Epoch 2, Loss 1.6177341938018799
Epoch 2, Loss 1.3591456413269043
Epoch 2, Loss 1.3552546501159668
Epoch 2, Loss 1.700721025466919
Epoch 2, Loss 1.5025629997253418
Epoch 2, Loss 1.7380791902542114
Epoch 2, Loss 1.4552404880523682
Epoch 2, Loss 1.4073638916015625
Epoch 2, Loss 1.685255765914917
Epoch 2, Loss 1.2512426376342773
Epoch 2, Loss 1.4989361763000488
Epoch 2, Loss 1.4337191581726074
Epoch 2, Loss 1.9238563776016235
Epoch 2, Loss 1.5253225564956665
Epoch 2, Loss 1.4192780256271362
Epoch 2, Loss 1.3466325998306274
Epoch 2, Loss 1.2324621677398682
Epoch 2, Loss 1.3456989526748657
Epoch 2, Loss 1.5006344318389893
Epoch 2, Loss 1.592153787612915
Epoch 2, Loss 1.3227531909942627
Epoch 2, Loss 1.3968530893325806
Epoch 2, Loss 1.2369649410247803
Epoch 2, Loss 1.4970403909683228
Epoch 2, Loss 1.4391765594482422
Epoch 2, Loss 1.6369444131851196
Epoch 2, Loss 1.414367914199829
Epoch 2, Loss 1.7419261932373047
Epoch 2, Loss 1.4557058811187744
Epoch 2, Loss 1.2599941492080688
Epoch 2, Loss 1.731783390045166
Epoch 2, Loss 1.608886957168579
Epoch 2, Loss 1.2529417276382446
Epoch 2, Loss 1.3802416324615479
Epoch 2, Loss 1.4927818775177002
Epoch 2, Loss 1.5472095012664795
Epoch 2, Loss 1.5669280290603638
Epoch 2, Loss 1.5150359869003296
Epoch 2, Loss 1.6746834516525269
Epoch 2, Loss 1.0480552911758423
Epoch 2, Loss 1.5535739660263062
Epoch 2, Loss 1.5265779495239258
Epoch 2, Loss 1.3840503692626953
Epoch 2, Loss 1.7055798768997192
Epoch 2, Loss 1.5464352369308472
Epoch 2, Loss 1.174098014831543
Epoch 2, Loss 1.6471370458602905
Epoch 2, Loss 1.2784475088119507
Epoch 2, Loss 1.055302381515503
Epoch 2, Loss 1.3225092887878418
Epoch 2, Loss 1.291968584060669
Epoch 2, Loss 1.1358336210250854
Epoch 2, Loss 1.5210617780685425
Epoch 2, Loss 1.2005878686904907
Epoch 2, Loss 1.7784743309020996
Epoch 2, Loss 1.112040400505066
Epoch 2, Loss 1.4387006759643555
Epoch 2, Loss 1.2630209922790527
Epoch 2, Loss 1.3373013734817505
Epoch 2, Loss 1.2956228256225586
Epoch 2, Loss 1.4301007986068726
Epoch 2, Loss 1.5055773258209229
Epoch 2, Loss 1.3266152143478394
Epoch 2, Loss 1.4556552171707153
Epoch 2, Loss 1.2938069105148315
Epoch 3, Loss 1.2657749652862549
Epoch 3, Loss 1.3685088157653809
Epoch 3, Loss 1.3917596340179443
Epoch 3, Loss 1.454487681388855
Epoch 3, Loss 1.2210859060287476
Epoch 3, Loss 1.181883692741394
Epoch 3, Loss 1.4901630878448486
Epoch 3, Loss 1.639451026916504
Epoch 3, Loss 1.2693920135498047
Epoch 3, Loss 1.4349827766418457
Epoch 3, Loss 1.2711045742034912
Epoch 3, Loss 1.339880347251892
Epoch 3, Loss 1.418124794960022
Epoch 3, Loss 1.4467421770095825
Epoch 3, Loss 1.444687843322754
Epoch 3, Loss 1.1931383609771729
Epoch 3, Loss 1.4755314588546753
Epoch 3, Loss 1.1279876232147217
Epoch 3, Loss 1.3665730953216553
Epoch 3, Loss 1.8242483139038086
Epoch 3, Loss 1.568402647972107
Epoch 3, Loss 1.4150508642196655
Epoch 3, Loss 1.6502350568771362
Epoch 3, Loss 1.5290955305099487
Epoch 3, Loss 1.2656736373901367
Epoch 3, Loss 1.5326241254806519
Epoch 3, Loss 1.5800015926361084
Epoch 3, Loss 1.3273322582244873
Epoch 3, Loss 1.6606097221374512
Epoch 3, Loss 1.1275347471237183
Epoch 3, Loss 1.5929442644119263
Epoch 3, Loss 1.332479476928711
Epoch 3, Loss 1.6427674293518066
Epoch 3, Loss 1.2133643627166748
Epoch 3, Loss 1.4062793254852295
Epoch 3, Loss 1.4059674739837646
Epoch 3, Loss 1.4475295543670654
Epoch 3, Loss 1.4280588626861572
Epoch 3, Loss 1.7738473415374756
Epoch 3, Loss 1.3956644535064697
Epoch 3, Loss 1.5311213731765747
Epoch 3, Loss 0.9934478998184204
Epoch 3, Loss 1.545434832572937
Epoch 3, Loss 1.4548133611679077
Epoch 3, Loss 1.0867449045181274
Epoch 3, Loss 1.391006588935852
Epoch 3, Loss 1.4315217733383179
Epoch 3, Loss 1.6717965602874756
Epoch 3, Loss 1.5112428665161133
Epoch 3, Loss 1.488818883895874
Epoch 3, Loss 1.6172558069229126
Epoch 3, Loss 1.2331808805465698
Epoch 3, Loss 1.4733998775482178
Epoch 3, Loss 1.3625138998031616
Epoch 3, Loss 1.4386574029922485
Epoch 3, Loss 1.3568108081817627
Epoch 3, Loss 1.258704423904419
Epoch 3, Loss 1.6578422784805298
Epoch 3, Loss 1.4115132093429565
Epoch 3, Loss 1.2691428661346436
Epoch 3, Loss 1.4916046857833862
Epoch 3, Loss 1.530895709991455
Epoch 3, Loss 1.6672450304031372
Epoch 3, Loss 1.6978107690811157
Epoch 3, Loss 1.5214213132858276
Epoch 3, Loss 1.4066779613494873
Epoch 3, Loss 1.7450149059295654
Epoch 3, Loss 1.5878835916519165
Epoch 3, Loss 1.5820975303649902
Epoch 3, Loss 1.3010077476501465
Epoch 3, Loss 1.6103583574295044
Epoch 3, Loss 1.2728753089904785
Epoch 3, Loss 1.5998402833938599
Epoch 3, Loss 1.3423200845718384
Epoch 3, Loss 1.3012176752090454
Epoch 3, Loss 1.3549573421478271
Epoch 3, Loss 1.2880171537399292
Epoch 3, Loss 1.4422558546066284
Epoch 3, Loss 1.3892611265182495
Epoch 3, Loss 1.8734943866729736
Epoch 3, Loss 1.5323463678359985
Epoch 3, Loss 1.2365837097167969
Epoch 3, Loss 1.3567500114440918
Epoch 3, Loss 1.1825237274169922
Epoch 3, Loss 1.4575679302215576
Epoch 3, Loss 1.44956636428833
Epoch 3, Loss 1.3784416913986206
Epoch 3, Loss 1.6067789793014526
Epoch 3, Loss 1.191642165184021
Epoch 3, Loss 1.6875654458999634
Epoch 3, Loss 1.5299079418182373
Epoch 3, Loss 1.601672887802124
Epoch 3, Loss 1.446716070175171
Epoch 3, Loss 1.482204556465149
Epoch 3, Loss 1.5481945276260376
Epoch 3, Loss 1.641560435295105
Epoch 3, Loss 1.1678754091262817
Epoch 3, Loss 1.2199982404708862
Epoch 3, Loss 1.5881229639053345
Epoch 3, Loss 1.4570183753967285
Epoch 3, Loss 1.6045030355453491
Epoch 3, Loss 1.8124254941940308
Epoch 3, Loss 1.1126610040664673
Epoch 3, Loss 1.514694333076477
Epoch 3, Loss 1.3265814781188965
Epoch 3, Loss 1.3347218036651611
Epoch 3, Loss 1.184769868850708
Epoch 3, Loss 1.6264007091522217
Epoch 3, Loss 1.6782984733581543
Epoch 3, Loss 1.3247326612472534
Epoch 3, Loss 1.570529818534851
Epoch 3, Loss 1.5758389234542847
Epoch 3, Loss 1.3637256622314453
Epoch 3, Loss 1.2797119617462158
Epoch 3, Loss 1.4374923706054688
Epoch 3, Loss 1.4566384553909302
Epoch 3, Loss 1.2352689504623413
Epoch 3, Loss 1.6088505983352661
Epoch 3, Loss 1.5399131774902344
Epoch 3, Loss 1.5017908811569214
Epoch 3, Loss 1.657196044921875
Epoch 3, Loss 1.3789411783218384
Epoch 3, Loss 1.2777636051177979
Epoch 3, Loss 1.6071970462799072
Epoch 3, Loss 1.6217520236968994
Epoch 3, Loss 1.2750431299209595
Epoch 3, Loss 1.499721646308899
Epoch 3, Loss 1.741183876991272
Epoch 3, Loss 1.2842175960540771
Epoch 3, Loss 2.0048739910125732
Epoch 3, Loss 1.4397671222686768
Epoch 3, Loss 1.41280996799469
Epoch 3, Loss 1.327250361442566
Epoch 3, Loss 1.3460841178894043
Epoch 3, Loss 1.3374409675598145
Epoch 3, Loss 1.3694055080413818
Epoch 3, Loss 1.6188303232192993
Epoch 3, Loss 1.5583730936050415
Epoch 3, Loss 1.1917520761489868
Epoch 3, Loss 1.3353607654571533
Epoch 3, Loss 1.2896196842193604
Epoch 3, Loss 1.3106706142425537
Epoch 3, Loss 1.3984615802764893
Epoch 3, Loss 1.4131144285202026
Epoch 3, Loss 1.7911815643310547
Epoch 3, Loss 1.4342396259307861
Epoch 3, Loss 1.2494813203811646
Epoch 3, Loss 1.3417868614196777
Epoch 3, Loss 1.1329594850540161
Epoch 3, Loss 1.3146885633468628
Epoch 3, Loss 1.1411229372024536
Epoch 3, Loss 1.4046076536178589
Epoch 3, Loss 1.422649621963501
Epoch 3, Loss 1.3979917764663696
Epoch 3, Loss 1.4007906913757324
Epoch 3, Loss 1.742323398590088
Epoch 3, Loss 1.334319829940796
Epoch 3, Loss 1.505510926246643
Epoch 3, Loss 1.5146727561950684
Epoch 3, Loss 1.3049325942993164
Epoch 3, Loss 1.5551166534423828
Epoch 3, Loss 1.5730377435684204
Epoch 3, Loss 1.4128203392028809
Epoch 3, Loss 1.3167755603790283
Epoch 3, Loss 1.518812656402588
Epoch 3, Loss 1.5243498086929321
Epoch 3, Loss 1.3108272552490234
Epoch 3, Loss 1.486115574836731
Epoch 3, Loss 1.165726900100708
Epoch 3, Loss 1.255001425743103
Epoch 3, Loss 1.2435463666915894
Epoch 3, Loss 1.4704406261444092
Epoch 3, Loss 1.3664522171020508
Epoch 3, Loss 1.4193373918533325
Epoch 3, Loss 1.3270068168640137
Epoch 3, Loss 1.2870937585830688
Epoch 3, Loss 1.6283689737319946
Epoch 3, Loss 1.6018389463424683
Epoch 3, Loss 1.1878634691238403
Epoch 3, Loss 1.181381344795227
Epoch 3, Loss 1.3414957523345947
Epoch 3, Loss 1.5913832187652588
Epoch 3, Loss 1.2617591619491577
Epoch 3, Loss 1.276902198791504
Epoch 3, Loss 1.2591112852096558
Epoch 3, Loss 1.5003361701965332
Epoch 3, Loss 1.6098048686981201
Epoch 3, Loss 1.3043162822723389
Epoch 3, Loss 1.4671623706817627
Epoch 3, Loss 1.2743574380874634
Epoch 3, Loss 1.333449363708496
Epoch 3, Loss 1.0447195768356323
Epoch 3, Loss 1.7496633529663086
Epoch 3, Loss 1.4349948167800903
Epoch 3, Loss 0.9941039681434631
Epoch 3, Loss 1.1381226778030396
Epoch 3, Loss 1.1700595617294312
Epoch 3, Loss 1.3808567523956299
Epoch 3, Loss 1.4743382930755615
Epoch 3, Loss 1.5828357934951782
Epoch 3, Loss 1.3578652143478394
Epoch 3, Loss 1.253328561782837
Epoch 3, Loss 1.6470190286636353
Epoch 3, Loss 1.3152893781661987
Epoch 3, Loss 1.3049588203430176
Epoch 3, Loss 1.75481116771698
Epoch 3, Loss 1.226052165031433
Epoch 3, Loss 1.434431552886963
Epoch 3, Loss 1.576608419418335
Epoch 3, Loss 1.5275124311447144
Epoch 3, Loss 1.2385188341140747
Epoch 3, Loss 1.1679117679595947
Epoch 3, Loss 1.4889219999313354
Epoch 3, Loss 1.3882681131362915
Epoch 3, Loss 1.471296787261963
Epoch 3, Loss 1.022072196006775
Epoch 3, Loss 1.1666314601898193
Epoch 3, Loss 1.365638017654419
Epoch 3, Loss 1.4819036722183228
Epoch 3, Loss 1.7000688314437866
Epoch 3, Loss 1.6323959827423096
Epoch 3, Loss 1.3997470140457153
Epoch 3, Loss 1.5760842561721802
Epoch 3, Loss 1.2480628490447998
Epoch 3, Loss 1.4412319660186768
Epoch 3, Loss 1.1057498455047607
Epoch 3, Loss 1.4089657068252563
Epoch 3, Loss 1.2607370615005493
Epoch 3, Loss 1.3391801118850708
Epoch 3, Loss 1.1781189441680908
Epoch 3, Loss 1.7425432205200195
Epoch 3, Loss 1.4165338277816772
Epoch 3, Loss 1.8061004877090454
Epoch 3, Loss 1.6210708618164062
Epoch 3, Loss 1.6151728630065918
Epoch 3, Loss 1.3837578296661377
Epoch 3, Loss 1.4376273155212402
Epoch 3, Loss 1.4341416358947754
Epoch 3, Loss 1.6254396438598633
Epoch 3, Loss 1.1672800779342651
Epoch 3, Loss 1.523203730583191
Epoch 3, Loss 1.3754456043243408
Epoch 3, Loss 1.3214695453643799
Epoch 3, Loss 0.9381955862045288
Epoch 3, Loss 1.6798808574676514
Epoch 3, Loss 1.4536024332046509
Epoch 3, Loss 1.3441954851150513
Epoch 3, Loss 1.3261338472366333
Epoch 3, Loss 1.459885597229004
Epoch 3, Loss 1.2846229076385498
Epoch 3, Loss 1.5398207902908325
Epoch 3, Loss 0.9090222120285034
Epoch 3, Loss 1.3650181293487549
Epoch 3, Loss 1.2454298734664917
Epoch 3, Loss 1.3336719274520874
Epoch 3, Loss 1.4035537242889404
Epoch 3, Loss 0.9969670176506042
Epoch 3, Loss 1.3733644485473633
Epoch 3, Loss 1.6806672811508179
Epoch 3, Loss 1.1753177642822266
Epoch 3, Loss 1.397196888923645
Epoch 3, Loss 1.438379168510437
Epoch 3, Loss 1.463959813117981
Epoch 3, Loss 1.4594169855117798
Epoch 3, Loss 1.2245839834213257
Epoch 3, Loss 1.248319387435913
Epoch 3, Loss 1.325327754020691
Epoch 3, Loss 1.6232614517211914
Epoch 3, Loss 1.192349910736084
Epoch 3, Loss 1.3080886602401733
Epoch 3, Loss 1.5868333578109741
Epoch 3, Loss 1.5101203918457031
Epoch 3, Loss 1.6002157926559448
Epoch 3, Loss 1.4421924352645874
Epoch 3, Loss 1.438562273979187
Epoch 3, Loss 1.22853422164917
Epoch 3, Loss 1.102024793624878
Epoch 3, Loss 1.0905317068099976
Epoch 3, Loss 1.378516435623169
Epoch 3, Loss 1.1065865755081177
Epoch 3, Loss 1.2351804971694946
Epoch 3, Loss 1.1582094430923462
Epoch 3, Loss 1.4239323139190674
Epoch 3, Loss 1.0838066339492798
Epoch 3, Loss 1.4736006259918213
Epoch 3, Loss 1.2326642274856567
Epoch 3, Loss 1.7137911319732666
Epoch 3, Loss 1.691973328590393
Epoch 3, Loss 1.3055859804153442
Epoch 3, Loss 1.0706970691680908
Epoch 3, Loss 1.3448677062988281
Epoch 3, Loss 1.5532619953155518
Epoch 3, Loss 1.263440489768982
Epoch 3, Loss 1.306041955947876
Epoch 3, Loss 1.678679347038269
Epoch 3, Loss 1.1568427085876465
Epoch 3, Loss 1.1706748008728027
Epoch 3, Loss 1.1872162818908691
Epoch 3, Loss 1.2646400928497314
Epoch 3, Loss 0.9924498200416565
Epoch 3, Loss 1.5050513744354248
Epoch 3, Loss 1.4457910060882568
Epoch 3, Loss 1.3851691484451294
Epoch 3, Loss 1.4667108058929443
Epoch 3, Loss 1.6174603700637817
Epoch 3, Loss 1.507022738456726
Epoch 3, Loss 1.4685307741165161
Epoch 3, Loss 0.9865709543228149
Epoch 3, Loss 1.5189127922058105
Epoch 3, Loss 1.8143751621246338
Epoch 3, Loss 1.5179133415222168
Epoch 3, Loss 1.0695254802703857
Epoch 3, Loss 1.039542555809021