import numpy as np
import re
import pickle
import os
import glob
import pandas as pd
from google.colab import drive
import cv2
from PIL import Image
from pathlib import Path
import fastai
from fastai.vision import *
from fastai.vision.models import *
import torch
import xgboost as xgb

fastai.__version__

'1.0.60'
drive.mount('/content/drive')

Go to this URL in a browser: https://accounts.google.com/o/oauth2/auth?client_id=947318989803-6bn6qk8qdgf4n4g3pfee6491hc0brc4i.apps.googleusercontent.com&redirect_uri=urn%3aietf%3awg%3aoauth%3a2.0%3aoob&response_type=code&scope=email%20https%3a%2f%2fwww.googleapis.com%2fauth%2fdocs.test%20https%3a%2f%2fwww.googleapis.com%2fauth%2fdrive%20https%3a%2f%2fwww.googleapis.com%2fauth%2fdrive.photos.readonly%20https%3a%2f%2fwww.googleapis.com%2fauth%2fpeopleapi.readonly

Enter your authorization code:
··········
Mounted at /content/drive

os.chdir('/content/drive/My Drive/ICLR')

healthy_wheat=glob.glob('/content/drive/My Drive/ICLR/ICLR/train/train/healthy_wheat/*.*')
leaf_rust=glob.glob('/content/drive/My Drive/ICLR/ICLR/train/train/leaf_rust/*.*')
stem_rust=glob.glob('/content/drive/My Drive/ICLR/ICLR/train/train/stem_rust/*.*')

img_path=[]
img_path=img_path+healthy_wheat+leaf_rust+stem_rust
df=pd.DataFrame({'img_path':img_path})
df.shape

(1189, 1)
df.head()
img_path
0 /content/drive/My Drive/ICLR/ICLR/train/train/...
1 /content/drive/My Drive/ICLR/ICLR/train/train/...
2 /content/drive/My Drive/ICLR/ICLR/train/train/...
3 /content/drive/My Drive/ICLR/ICLR/train/train/...
4 /content/drive/My Drive/ICLR/ICLR/train/train/... 
df['img_path'] = df['img_path'].apply(lambda x: re.sub("/","",x,1))
df.head()
img_path
0 content/drive/My Drive/ICLR/ICLR/train/train/h...
1 content/drive/My Drive/ICLR/ICLR/train/train/h...
2 content/drive/My Drive/ICLR/ICLR/train/train/h...
3 content/drive/My Drive/ICLR/ICLR/train/train/h...
4 content/drive/My Drive/ICLR/ICLR/train/train/h... 
df['disease_class'] = df['img_path'].apply(lambda x: x.split("/")[7])
df['ext'] = df['img_path'].apply(lambda x: x.split(".")[-1])
df.head()
img_path disease_class ext
0 content/drive/My Drive/ICLR/ICLR/train/train/h... healthy_wheat jpg
1 content/drive/My Drive/ICLR/ICLR/train/train/h... healthy_wheat jpg
2 content/drive/My Drive/ICLR/ICLR/train/train/h... healthy_wheat jpg
3 content/drive/My Drive/ICLR/ICLR/train/train/h... healthy_wheat jfif
4 content/drive/My Drive/ICLR/ICLR/train/train/h... healthy_wheat jpg 
df=df[df.ext!='jfif']
df.drop('ext',axis=1,inplace=True)
df.head()
img_path disease_class
0 content/drive/My Drive/ICLR/ICLR/train/train/h... healthy_wheat
1 content/drive/My Drive/ICLR/ICLR/train/train/h... healthy_wheat
2 content/drive/My Drive/ICLR/ICLR/train/train/h... healthy_wheat
4 content/drive/My Drive/ICLR/ICLR/train/train/h... healthy_wheat
5 content/drive/My Drive/ICLR/ICLR/train/train/h... healthy_wheat 
df.shape

(876, 2)
df.reset_index(inplace=True)

df.drop('index',axis=1,inplace=True)

df.head()
img_path disease_class
0 content/drive/My Drive/ICLR/ICLR/train/train/h... healthy_wheat
1 content/drive/My Drive/ICLR/ICLR/train/train/h... healthy_wheat
2 content/drive/My Drive/ICLR/ICLR/train/train/h... healthy_wheat
3 content/drive/My Drive/ICLR/ICLR/train/train/h... healthy_wheat
4 content/drive/My Drive/ICLR/ICLR/train/train/h... healthy_wheat 
test_img_path=glob.glob('/content/drive/My Drive/ICLR/ICLR/test/test/*.*')
test_df=pd.DataFrame({'test_img_path':test_img_path})
test_df['ext'] = test_df['test_img_path'].apply(lambda x: x.split(".")[-1])
test_df=test_df[test_df.ext!='jfif']
test_df.shape

(610, 2)
test_df['test_img_path'] = test_df['test_img_path'].apply(lambda x: re.sub("/","",x,1))
test_df.drop('ext',axis=1,inplace=True)
test_df.reset_index(inplace=True)
test_df.drop('index',axis=1,inplace=True)
test_df.head()
test_img_path
0 content/drive/My Drive/ICLR/ICLR/test/test/008...
1 content/drive/My Drive/ICLR/ICLR/test/test/085...
2 content/drive/My Drive/ICLR/ICLR/test/test/00A...
3 content/drive/My Drive/ICLR/ICLR/test/test/01O...
4 content/drive/My Drive/ICLR/ICLR/test/test/0K9... 
df['disease_class'].value_counts()

stem_rust        376
leaf_rust        358
healthy_wheat    142
Name: disease_class, dtype: int64
from sklearn.model_selection import train_test_split
train_names = df.index.values
train_labels = np.asarray(df['disease_class'].values)
tr_n, tr_idx, val_n, val_idx = train_test_split(train_names, range(len(train_names)), test_size=0.1, stratify=train_labels, random_state=123)

from tqdm.notebook import tqdm
count=0
total=np.zeros(3)
total_2=np.zeros(3)
for i,fn in tqdm(enumerate(df['img_path']),'computing statistics...'):
  path = os.path.join('/', fn)
  imagearray = np.array(PIL.Image.open(os.path.join('/',fn)))/255
  if np.array(imagearray).shape[-1]==3:
    total += np.array(imagearray).reshape(-1,3).mean(axis=0)
    total_2 += ((np.array(imagearray).reshape(-1,3))**2).mean(axis=0)
    count+=1
channel_avr = total/count
channel_std = np.sqrt(total_2/count - channel_avr**2)
channel_avr,channel_std















(array([0.454764, 0.475819, 0.321528]), array([0.232232, 0.218677, 0.219719]))














    

    
    







    


imagenet_stats



    



















([0.485, 0.456, 0.406], [0.229, 0.224, 0.225])














    

    
    







    


def create_databunch(val_idx,sz,bs):
  test = ImageList.from_df(test_df,path='/')
  data = (ImageList.from_df(df, path='/')
              .split_by_idx(val_idx)
              .label_from_df(cols='disease_class')
              .add_test(test)
              .transform(get_transforms(flip_vert=True), size=sz)
              .databunch(path='.', bs=bs)
              .normalize(imagenet_stats)
            )
  return data



    








    

    
    







    


arch=models.densenet201
sz=256
bs=32



    








    

    
    







    


import torch
import gc
!torch.cuda.empty_cache()
learner.destroy()
gc.collect()



    

















/bin/bash: -c: line 1: syntax error: unexpected end of file
this Learner object self-destroyed - it still exists, but no longer usable















0














    

    
    







    


data = create_databunch(val_idx,sz,bs)
data.show_batch(rows=2, figsize=(4,4))

def getLearner():
    return cnn_learner(data, arch, metrics=[accuracy]).mixup()



    
































    

    
    







    


lrs = []
losses = []
wds = []
iter_count = 600

learner = getLearner() 
# WEIGHT DECAY = 1e-6
learner.lr_find(wd=1e-6,num_it=iter_count)
lrs.append(learner.recorder.lrs)
losses.append(learner.recorder.losses)
wds.append('1e-6')
learner = getLearner() #reset learner - this gets more consistent starting conditions

# WEIGHT DECAY = 1e-4
learner.lr_find(wd=1e-4, num_it=iter_count)
lrs.append(learner.recorder.lrs)
losses.append(learner.recorder.losses)
wds.append('1e-4')
learner = getLearner() #reset learner - this gets more consistent starting conditions

# WEIGHT DECAY = 1e-2
learner.lr_find(wd=1e-2, num_it=iter_count)
lrs.append(learner.recorder.lrs)
losses.append(learner.recorder.losses)
wds.append('1e-2')
learner = getLearner() #reset learner



    


















    

        
      <progress value='20' class='' max='25', style='width:300px; height:20px; vertical-align: middle;'></progress>
      80.00% [20/25 48:18<12:04]
    

    


  
    

      epoch
      train_loss
      valid_loss
      accuracy
      time
    

  
  
    

      0
      2.011008
      #na#
      04:10
    

    

      1
      2.028271
      #na#
      02:19
    

    

      2
      2.018274
      #na#
      02:18
    

    

      3
      1.969068
      #na#
      02:18
    

    

      4
      1.955951
      #na#
      02:20
    

    

      5
      1.938861
      #na#
      02:20
    

    

      6
      1.939544
      #na#
      02:18
    

    

      7
      1.856004
      #na#
      02:18
    

    

      8
      1.738340
      #na#
      02:17
    

    

      9
      1.559586
      #na#
      02:18
    

    

      10
      1.438459
      #na#
      02:18
    

    

      11
      1.317430
      #na#
      02:21
    

    

      12
      1.246319
      #na#
      02:18
    

    

      13
      1.252726
      #na#
      02:21
    

    

      14
      1.133312
      #na#
      02:21
    

    

      15
      1.024948
      #na#
      02:19
    

    

      16
      1.052209
      #na#
      02:20
    

    

      17
      1.194508
      #na#
      02:16
    

    

      18
      2.130605
      #na#
      02:20
    

    

      19
      2.944213
      #na#
      02:19
    

  



    

        
      <progress value='7' class='' max='24', style='width:300px; height:20px; vertical-align: middle;'></progress>
      29.17% [7/24 00:49<01:59 3.9260]
    

    
</div>

</div>







LR Finder is complete, type {learner_name}.recorder.plot() to see the graph.














    

        
      <progress value='20' class='' max='25', style='width:300px; height:20px; vertical-align: middle;'></progress>
      80.00% [20/25 47:01<11:45]
    

    


  
    

      epoch
      train_loss
      valid_loss
      accuracy
      time
    

  
  
    

      0
      1.974997
      #na#
      02:18
    

    

      1
      2.010628
      #na#
      02:19
    

    

      2
      1.951568
      #na#
      02:18
    

    

      3
      1.928268
      #na#
      02:19
    

    

      4
      1.882828
      #na#
      02:22
    

    

      5
      1.879035
      #na#
      02:22
    

    

      6
      1.841328
      #na#
      02:26
    

    

      7
      1.813223
      #na#
      02:21
    

    

      8
      1.669109
      #na#
      02:22
    

    

      9
      1.571100
      #na#
      02:20
    

    

      10
      1.461533
      #na#
      02:20
    

    

      11
      1.336195
      #na#
      02:21
    

    

      12
      1.246948
      #na#
      02:19
    

    

      13
      1.187644
      #na#
      02:23
    

    

      14
      1.085130
      #na#
      02:20
    

    

      15
      0.996098
      #na#
      02:19
    

    

      16
      1.018888
      #na#
      02:19
    

    

      17
      1.212861
      #na#
      02:19
    

    

      18
      1.893726
      #na#
      02:22
    

    

      19
      2.995594
      #na#
      02:22
    

  



    

        
      <progress value='6' class='' max='24', style='width:300px; height:20px; vertical-align: middle;'></progress>
      25.00% [6/24 00:37<01:52 3.8755]
    

    
</div>

</div>







LR Finder is complete, type {learner_name}.recorder.plot() to see the graph.














    

        
      <progress value='20' class='' max='25', style='width:300px; height:20px; vertical-align: middle;'></progress>
      80.00% [20/25 46:57<11:44]
    

    


  
    

      epoch
      train_loss
      valid_loss
      accuracy
      time
    

  
  
    

      0
      2.084592
      #na#
      02:19
    

    

      1
      2.144779
      #na#
      02:19
    

    

      2
      2.063327
      #na#
      02:18
    

    

      3
      2.015883
      #na#
      02:19
    

    

      4
      2.040844
      #na#
      02:20
    

    

      5
      2.044003
      #na#
      02:20
    

    

      6
      1.944221
      #na#
      02:20
    

    

      7
      1.816331
      #na#
      02:23
    

    

      8
      1.738291
      #na#
      02:20
    

    

      9
      1.622006
      #na#
      02:18
    

    

      10
      1.485781
      #na#
      02:19
    

    

      11
      1.367897
      #na#
      02:26
    

    

      12
      1.287109
      #na#
      02:22
    

    

      13
      1.246014
      #na#
      02:20
    

    

      14
      1.123771
      #na#
      02:18
    

    

      15
      1.005649
      #na#
      02:23
    

    

      16
      1.066494
      #na#
      02:20
    

    

      17
      1.277751
      #na#
      02:20
    

    

      18
      1.814309
      #na#
      02:21
    

    

      19
      2.669663
      #na#
      02:21
    

  



    

        
      <progress value='12' class='' max='24', style='width:300px; height:20px; vertical-align: middle;'></progress>
      50.00% [12/24 01:15<01:15 3.9202]
    

    
</div>

</div>







LR Finder is complete, type {learner_name}.recorder.plot() to see the graph.







</div>
</div>

</div>
    

    
    







    


# Plot weight decays
_, ax = plt.subplots(1,1)
min_y = 4.00
max_y = 5.55
for i in range(len(losses)):
    ax.plot(lrs[i], losses[i])
    min_y = min(np.asarray(losses[i]).min(), min_y)
ax.set_ylabel("Loss")
ax.set_xlabel("Learning Rate")
ax.set_xscale('log')
#ax ranges may need some tuning with different model architectures 
ax.set_xlim((1e-3,3e-1))
ax.set_ylim((min_y - 0.02,max_y))
ax.legend(wds)
ax.xaxis.set_major_formatter(plt.FormatStrFormatter('%.0e'))



    
































    

    
    







    


max_lr = 2e-2
wd = 1e-4
# 1cycle policy
learner.fit_one_cycle(cyc_len=8, max_lr=max_lr, wd=wd)



    



















  
    

      epoch
      train_loss
      valid_loss
      accuracy
      time
    

  
  
    

      0
      1.798677
      1.290595
      0.590909
      03:20
    

    

      1
      1.402241
      0.830046
      0.693182
      02:38
    

    

      2
      1.289120
      1.225668
      0.715909
      02:35
    

    

      3
      1.222847
      3.080747
      0.693182
      02:37
    

    

      4
      1.160970
      1.084033
      0.727273
      02:38
    

    

      5
      1.042271
      0.522933
      0.806818
      02:38
    

    

      6
      0.929025
      0.438545
      0.852273
      02:39
    

    

      7
      0.852060
      0.438347
      0.852273
      02:39
    

  















    

    
    







    


learner.recorder.plot_lr()



    
































    

    
    







    


# and plot the losses of the first cycle
learner.recorder.plot_losses()



    
































    

    
    







    


# predict the validation set with our model
interp = ClassificationInterpretation.from_learner(learner)
interp.plot_confusion_matrix(title='Confusion matrix')



    













































    

    
    







    


learner.save("densenet201_model", return_path=True)



    



















PosixPath('models/densenet201_model.pth')














    

    
    







    


learner.load('/content/drive/My Drive/ICLR/models/densenet201_model')
# unfreeze and run learning rate finder again
learner.unfreeze()
learner.lr_find(wd=wd)

# plot learning rate finder results
learner.recorder.plot()



    


















    

        
      <progress value='3' class='' max='5', style='width:300px; height:20px; vertical-align: middle;'></progress>
      60.00% [3/5 07:07<04:44]
    

    


  
    

      epoch
      train_loss
      valid_loss
      accuracy
      time
    

  
  
    

      0
      0.714455
      #na#
      02:25
    

    

      1
      0.676535
      #na#
      02:22
    

    

      2
      0.979520
      #na#
      02:19
    

  



    

        
      <progress value='11' class='' max='24', style='width:300px; height:20px; vertical-align: middle;'></progress>
      45.83% [11/24 01:11<01:24 2.0032]
    

    
</div>

</div>







LR Finder is complete, type {learner_name}.recorder.plot() to see the graph.





















</div>
</div>

</div>
    

    
    







    


learner.fit_one_cycle(cyc_len=12, max_lr=slice(4e-5,4e-4))



    



















  
    

      epoch
      train_loss
      valid_loss
      accuracy
      time
    

  
  
    

      0
      0.686068
      0.434821
      0.863636
      02:37
    

    

      1
      0.672493
      0.429090
      0.840909
      02:39
    

    

      2
      0.639352
      0.389521
      0.897727
      02:40
    

    

      3
      0.664917
      0.440811
      0.863636
      02:41
    

    

      4
      0.660091
      0.419481
      0.875000
      02:40
    

    

      5
      0.633723
      0.362097
      0.909091
      02:41
    

    

      6
      0.629357
      0.439625
      0.863636
      02:40
    

    

      7
      0.610594
      0.447234
      0.875000
      02:41
    

    

      8
      0.596393
      0.404710
      0.886364
      02:40
    

    

      9
      0.571786
      0.403621
      0.886364
      02:38
    

    

      10
      0.577577
      0.394473
      0.886364
      02:37
    

    

      11
      0.562980
      0.396398
      0.886364
      02:40
    

  















    

    
    







    


learner.recorder.plot_losses()



    
































    

    
    







    


interp = ClassificationInterpretation.from_learner(learner)
interp.plot_confusion_matrix(title='Confusion matrix')



    













































    

    
    







    


learner.show_results(rows=5, figsize=(10,15))



    
































    

    
    







    


interp.plot_top_losses(9, figsize=(15,15))



    
































    

    
    







    


learner.save("densenet201_model"+"stage-2", return_path=True)



    



















PosixPath('models/densenet201_modelstage-2.pth')














    

    
    







    


learner.load('/content/drive/My Drive/ICLR/models/densenet201_modelstage-2')



    



















Learner(data=ImageDataBunch;

Train: LabelList (788 items)
x: ImageList
Image (3, 512, 512),Image (3, 512, 512),Image (3, 512, 512),Image (3, 512, 512),Image (3, 512, 512)
y: CategoryList
healthy_wheat,healthy_wheat,healthy_wheat,healthy_wheat,healthy_wheat
Path: /;

Valid: LabelList (88 items)
x: ImageList
Image (3, 512, 512),Image (3, 512, 512),Image (3, 512, 512),Image (3, 512, 512),Image (3, 512, 512)
y: CategoryList
stem_rust,stem_rust,leaf_rust,leaf_rust,leaf_rust
Path: /;

Test: LabelList (610 items)
x: ImageList
Image (3, 512, 512),Image (3, 512, 512),Image (3, 512, 512),Image (3, 512, 512),Image (3, 512, 512)
y: EmptyLabelList
,,,,
Path: /, model=Sequential(
  (0): Sequential(
    (0): Sequential(
      (conv0): Conv2d(3, 64, kernel_size=(7, 7), stride=(2, 2), padding=(3, 3), bias=False)
      (norm0): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
      (relu0): ReLU(inplace=True)
      (pool0): MaxPool2d(kernel_size=3, stride=2, padding=1, dilation=1, ceil_mode=False)
      (denseblock1): _DenseBlock(
        (denselayer1): _DenseLayer(
          (norm1): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (relu1): ReLU(inplace=True)
          (conv1): Conv2d(64, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (norm2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (relu2): ReLU(inplace=True)
          (conv2): Conv2d(128, 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
        )
        (denselayer2): _DenseLayer(
          (norm1): BatchNorm2d(96, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (relu1): ReLU(inplace=True)
          (conv1): Conv2d(96, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (norm2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (relu2): ReLU(inplace=True)
          (conv2): Conv2d(128, 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
        )
        (denselayer3): _DenseLayer(
          (norm1): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (relu1): ReLU(inplace=True)
          (conv1): Conv2d(128, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (norm2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (relu2): ReLU(inplace=True)
          (conv2): Conv2d(128, 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
        )
        (denselayer4): _DenseLayer(
          (norm1): BatchNorm2d(160, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (relu1): ReLU(inplace=True)
          (conv1): Conv2d(160, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (norm2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (relu2): ReLU(inplace=True)
          (conv2): Conv2d(128, 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
        )
        (denselayer5): _DenseLayer(
          (norm1): BatchNorm2d(192, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (relu1): ReLU(inplace=True)
          (conv1): Conv2d(192, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (norm2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (relu2): ReLU(inplace=True)
          (conv2): Conv2d(128, 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
        )
        (denselayer6): _DenseLayer(
          (norm1): BatchNorm2d(224, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (relu1): ReLU(inplace=True)
          (conv1): Conv2d(224, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (norm2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (relu2): ReLU(inplace=True)
          (conv2): Conv2d(128, 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
        )
      )
      (transition1): _Transition(
        (norm): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
        (relu): ReLU(inplace=True)
        (conv): Conv2d(256, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)
        (pool): AvgPool2d(kernel_size=2, stride=2, padding=0)
      )
      (denseblock2): _DenseBlock(
        (denselayer1): _DenseLayer(
          (norm1): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (relu1): ReLU(inplace=True)
          (conv1): Conv2d(128, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (norm2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (relu2): ReLU(inplace=True)
          (conv2): Conv2d(128, 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
        )
        (denselayer2): _DenseLayer(
          (norm1): BatchNorm2d(160, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (relu1): ReLU(inplace=True)
          (conv1): Conv2d(160, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (norm2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (relu2): ReLU(inplace=True)
          (conv2): Conv2d(128, 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
        )
        (denselayer3): _DenseLayer(
          (norm1): BatchNorm2d(192, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (relu1): ReLU(inplace=True)
          (conv1): Conv2d(192, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (norm2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (relu2): ReLU(inplace=True)
          (conv2): Conv2d(128, 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
        )
        (denselayer4): _DenseLayer(
          (norm1): BatchNorm2d(224, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (relu1): ReLU(inplace=True)
          (conv1): Conv2d(224, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (norm2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (relu2): ReLU(inplace=True)
          (conv2): Conv2d(128, 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
        )
        (denselayer5): _DenseLayer(
          (norm1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (relu1): ReLU(inplace=True)
          (conv1): Conv2d(256, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (norm2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (relu2): ReLU(inplace=True)
          (conv2): Conv2d(128, 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
        )
        (denselayer6): _DenseLayer(
          (norm1): BatchNorm2d(288, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (relu1): ReLU(inplace=True)
          (conv1): Conv2d(288, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (norm2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (relu2): ReLU(inplace=True)
          (conv2): Conv2d(128, 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
        )
        (denselayer7): _DenseLayer(
          (norm1): BatchNorm2d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (relu1): ReLU(inplace=True)
          (conv1): Conv2d(320, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (norm2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (relu2): ReLU(inplace=True)
          (conv2): Conv2d(128, 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
        )
        (denselayer8): _DenseLayer(
          (norm1): BatchNorm2d(352, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (relu1): ReLU(inplace=True)
          (conv1): Conv2d(352, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (norm2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (relu2): ReLU(inplace=True)
          (conv2): Conv2d(128, 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
        )
        (denselayer9): _DenseLayer(
          (norm1): BatchNorm2d(384, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (relu1): ReLU(inplace=True)
          (conv1): Conv2d(384, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (norm2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (relu2): ReLU(inplace=True)
          (conv2): Conv2d(128, 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
        )
        (denselayer10): _DenseLayer(
          (norm1): BatchNorm2d(416, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (relu1): ReLU(inplace=True)
          (conv1): Conv2d(416, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (norm2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (relu2): ReLU(inplace=True)
          (conv2): Conv2d(128, 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
        )
        (denselayer11): _DenseLayer(
          (norm1): BatchNorm2d(448, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (relu1): ReLU(inplace=True)
          (conv1): Conv2d(448, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (norm2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (relu2): ReLU(inplace=True)
          (conv2): Conv2d(128, 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
        )
        (denselayer12): _DenseLayer(
          (norm1): BatchNorm2d(480, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (relu1): ReLU(inplace=True)
          (conv1): Conv2d(480, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (norm2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (relu2): ReLU(inplace=True)
          (conv2): Conv2d(128, 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
        )
      )
      (transition2): _Transition(
        (norm): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
        (relu): ReLU(inplace=True)
        (conv): Conv2d(512, 256, kernel_size=(1, 1), stride=(1, 1), bias=False)
        (pool): AvgPool2d(kernel_size=2, stride=2, padding=0)
      )
      (denseblock3): _DenseBlock(
        (denselayer1): _DenseLayer(
          (norm1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (relu1): ReLU(inplace=True)
          (conv1): Conv2d(256, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (norm2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (relu2): ReLU(inplace=True)
          (conv2): Conv2d(128, 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
        )
        (denselayer2): _DenseLayer(
          (norm1): BatchNorm2d(288, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (relu1): ReLU(inplace=True)
          (conv1): Conv2d(288, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (norm2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (relu2): ReLU(inplace=True)
          (conv2): Conv2d(128, 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
        )
        (denselayer3): _DenseLayer(
          (norm1): BatchNorm2d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (relu1): ReLU(inplace=True)
          (conv1): Conv2d(320, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (norm2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (relu2): ReLU(inplace=True)
          (conv2): Conv2d(128, 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
        )
        (denselayer4): _DenseLayer(
          (norm1): BatchNorm2d(352, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (relu1): ReLU(inplace=True)
          (conv1): Conv2d(352, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (norm2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (relu2): ReLU(inplace=True)
          (conv2): Conv2d(128, 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
        )
        (denselayer5): _DenseLayer(
          (norm1): BatchNorm2d(384, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (relu1): ReLU(inplace=True)
          (conv1): Conv2d(384, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (norm2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (relu2): ReLU(inplace=True)
          (conv2): Conv2d(128, 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
        )
        (denselayer6): _DenseLayer(
          (norm1): BatchNorm2d(416, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (relu1): ReLU(inplace=True)
          (conv1): Conv2d(416, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (norm2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (relu2): ReLU(inplace=True)
          (conv2): Conv2d(128, 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
        )
        (denselayer7): _DenseLayer(
          (norm1): BatchNorm2d(448, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (relu1): ReLU(inplace=True)
          (conv1): Conv2d(448, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (norm2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (relu2): ReLU(inplace=True)
          (conv2): Conv2d(128, 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
        )
        (denselayer8): _DenseLayer(
          (norm1): BatchNorm2d(480, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (relu1): ReLU(inplace=True)
          (conv1): Conv2d(480, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (norm2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (relu2): ReLU(inplace=True)
          (conv2): Conv2d(128, 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
        )
        (denselayer9): _DenseLayer(
          (norm1): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (relu1): ReLU(inplace=True)
          (conv1): Conv2d(512, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (norm2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (relu2): ReLU(inplace=True)
          (conv2): Conv2d(128, 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
        )
        (denselayer10): _DenseLayer(
          (norm1): BatchNorm2d(544, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (relu1): ReLU(inplace=True)
          (conv1): Conv2d(544, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (norm2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (relu2): ReLU(inplace=True)
          (conv2): Conv2d(128, 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
        )
        (denselayer11): _DenseLayer(
          (norm1): BatchNorm2d(576, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (relu1): ReLU(inplace=True)
          (conv1): Conv2d(576, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (norm2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (relu2): ReLU(inplace=True)
          (conv2): Conv2d(128, 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
        )
        (denselayer12): _DenseLayer(
          (norm1): BatchNorm2d(608, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (relu1): ReLU(inplace=True)
          (conv1): Conv2d(608, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (norm2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (relu2): ReLU(inplace=True)
          (conv2): Conv2d(128, 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
        )
        (denselayer13): _DenseLayer(
          (norm1): BatchNorm2d(640, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (relu1): ReLU(inplace=True)
          (conv1): Conv2d(640, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (norm2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (relu2): ReLU(inplace=True)
          (conv2): Conv2d(128, 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
        )
        (denselayer14): _DenseLayer(
          (norm1): BatchNorm2d(672, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (relu1): ReLU(inplace=True)
          (conv1): Conv2d(672, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (norm2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (relu2): ReLU(inplace=True)
          (conv2): Conv2d(128, 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
        )
        (denselayer15): _DenseLayer(
          (norm1): BatchNorm2d(704, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (relu1): ReLU(inplace=True)
          (conv1): Conv2d(704, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (norm2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (relu2): ReLU(inplace=True)
          (conv2): Conv2d(128, 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
        )
        (denselayer16): _DenseLayer(
          (norm1): BatchNorm2d(736, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (relu1): ReLU(inplace=True)
          (conv1): Conv2d(736, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (norm2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (relu2): ReLU(inplace=True)
          (conv2): Conv2d(128, 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
        )
        (denselayer17): _DenseLayer(
          (norm1): BatchNorm2d(768, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (relu1): ReLU(inplace=True)
          (conv1): Conv2d(768, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (norm2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (relu2): ReLU(inplace=True)
          (conv2): Conv2d(128, 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
        )
        (denselayer18): _DenseLayer(
          (norm1): BatchNorm2d(800, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (relu1): ReLU(inplace=True)
          (conv1): Conv2d(800, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (norm2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (relu2): ReLU(inplace=True)
          (conv2): Conv2d(128, 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
        )
        (denselayer19): _DenseLayer(
          (norm1): BatchNorm2d(832, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (relu1): ReLU(inplace=True)
          (conv1): Conv2d(832, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (norm2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (relu2): ReLU(inplace=True)
          (conv2): Conv2d(128, 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
        )
        (denselayer20): _DenseLayer(
          (norm1): BatchNorm2d(864, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (relu1): ReLU(inplace=True)
          (conv1): Conv2d(864, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (norm2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (relu2): ReLU(inplace=True)
          (conv2): Conv2d(128, 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
        )
        (denselayer21): _DenseLayer(
          (norm1): BatchNorm2d(896, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (relu1): ReLU(inplace=True)
          (conv1): Conv2d(896, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (norm2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (relu2): ReLU(inplace=True)
          (conv2): Conv2d(128, 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
        )
        (denselayer22): _DenseLayer(
          (norm1): BatchNorm2d(928, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (relu1): ReLU(inplace=True)
          (conv1): Conv2d(928, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (norm2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (relu2): ReLU(inplace=True)
          (conv2): Conv2d(128, 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
        )
        (denselayer23): _DenseLayer(
          (norm1): BatchNorm2d(960, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (relu1): ReLU(inplace=True)
          (conv1): Conv2d(960, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (norm2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (relu2): ReLU(inplace=True)
          (conv2): Conv2d(128, 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
        )
        (denselayer24): _DenseLayer(
          (norm1): BatchNorm2d(992, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (relu1): ReLU(inplace=True)
          (conv1): Conv2d(992, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (norm2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (relu2): ReLU(inplace=True)
          (conv2): Conv2d(128, 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
        )
        (denselayer25): _DenseLayer(
          (norm1): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (relu1): ReLU(inplace=True)
          (conv1): Conv2d(1024, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (norm2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (relu2): ReLU(inplace=True)
          (conv2): Conv2d(128, 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
        )
        (denselayer26): _DenseLayer(
          (norm1): BatchNorm2d(1056, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (relu1): ReLU(inplace=True)
          (conv1): Conv2d(1056, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (norm2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (relu2): ReLU(inplace=True)
          (conv2): Conv2d(128, 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
        )
        (denselayer27): _DenseLayer(
          (norm1): BatchNorm2d(1088, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (relu1): ReLU(inplace=True)
          (conv1): Conv2d(1088, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (norm2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (relu2): ReLU(inplace=True)
          (conv2): Conv2d(128, 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
        )
        (denselayer28): _DenseLayer(
          (norm1): BatchNorm2d(1120, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (relu1): ReLU(inplace=True)
          (conv1): Conv2d(1120, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (norm2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (relu2): ReLU(inplace=True)
          (conv2): Conv2d(128, 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
        )
        (denselayer29): _DenseLayer(
          (norm1): BatchNorm2d(1152, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (relu1): ReLU(inplace=True)
          (conv1): Conv2d(1152, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (norm2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (relu2): ReLU(inplace=True)
          (conv2): Conv2d(128, 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
        )
        (denselayer30): _DenseLayer(
          (norm1): BatchNorm2d(1184, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (relu1): ReLU(inplace=True)
          (conv1): Conv2d(1184, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (norm2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (relu2): ReLU(inplace=True)
          (conv2): Conv2d(128, 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
        )
        (denselayer31): _DenseLayer(
          (norm1): BatchNorm2d(1216, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (relu1): ReLU(inplace=True)
          (conv1): Conv2d(1216, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (norm2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (relu2): ReLU(inplace=True)
          (conv2): Conv2d(128, 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
        )
        (denselayer32): _DenseLayer(
          (norm1): BatchNorm2d(1248, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (relu1): ReLU(inplace=True)
          (conv1): Conv2d(1248, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (norm2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (relu2): ReLU(inplace=True)
          (conv2): Conv2d(128, 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
        )
        (denselayer33): _DenseLayer(
          (norm1): BatchNorm2d(1280, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (relu1): ReLU(inplace=True)
          (conv1): Conv2d(1280, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (norm2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (relu2): ReLU(inplace=True)
          (conv2): Conv2d(128, 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
        )
        (denselayer34): _DenseLayer(
          (norm1): BatchNorm2d(1312, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (relu1): ReLU(inplace=True)
          (conv1): Conv2d(1312, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (norm2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (relu2): ReLU(inplace=True)
          (conv2): Conv2d(128, 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
        )
        (denselayer35): _DenseLayer(
          (norm1): BatchNorm2d(1344, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (relu1): ReLU(inplace=True)
          (conv1): Conv2d(1344, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (norm2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (relu2): ReLU(inplace=True)
          (conv2): Conv2d(128, 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
        )
        (denselayer36): _DenseLayer(
          (norm1): BatchNorm2d(1376, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (relu1): ReLU(inplace=True)
          (conv1): Conv2d(1376, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (norm2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (relu2): ReLU(inplace=True)
          (conv2): Conv2d(128, 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
        )
        (denselayer37): _DenseLayer(
          (norm1): BatchNorm2d(1408, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (relu1): ReLU(inplace=True)
          (conv1): Conv2d(1408, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (norm2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (relu2): ReLU(inplace=True)
          (conv2): Conv2d(128, 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
        )
        (denselayer38): _DenseLayer(
          (norm1): BatchNorm2d(1440, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (relu1): ReLU(inplace=True)
          (conv1): Conv2d(1440, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (norm2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (relu2): ReLU(inplace=True)
          (conv2): Conv2d(128, 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
        )
        (denselayer39): _DenseLayer(
          (norm1): BatchNorm2d(1472, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (relu1): ReLU(inplace=True)
          (conv1): Conv2d(1472, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (norm2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (relu2): ReLU(inplace=True)
          (conv2): Conv2d(128, 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
        )
        (denselayer40): _DenseLayer(
          (norm1): BatchNorm2d(1504, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (relu1): ReLU(inplace=True)
          (conv1): Conv2d(1504, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (norm2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (relu2): ReLU(inplace=True)
          (conv2): Conv2d(128, 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
        )
        (denselayer41): _DenseLayer(
          (norm1): BatchNorm2d(1536, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (relu1): ReLU(inplace=True)
          (conv1): Conv2d(1536, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (norm2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (relu2): ReLU(inplace=True)
          (conv2): Conv2d(128, 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
        )
        (denselayer42): _DenseLayer(
          (norm1): BatchNorm2d(1568, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (relu1): ReLU(inplace=True)
          (conv1): Conv2d(1568, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (norm2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (relu2): ReLU(inplace=True)
          (conv2): Conv2d(128, 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
        )
        (denselayer43): _DenseLayer(
          (norm1): BatchNorm2d(1600, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (relu1): ReLU(inplace=True)
          (conv1): Conv2d(1600, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (norm2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (relu2): ReLU(inplace=True)
          (conv2): Conv2d(128, 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
        )
        (denselayer44): _DenseLayer(
          (norm1): BatchNorm2d(1632, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (relu1): ReLU(inplace=True)
          (conv1): Conv2d(1632, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (norm2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (relu2): ReLU(inplace=True)
          (conv2): Conv2d(128, 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
        )
        (denselayer45): _DenseLayer(
          (norm1): BatchNorm2d(1664, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (relu1): ReLU(inplace=True)
          (conv1): Conv2d(1664, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (norm2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (relu2): ReLU(inplace=True)
          (conv2): Conv2d(128, 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
        )
        (denselayer46): _DenseLayer(
          (norm1): BatchNorm2d(1696, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (relu1): ReLU(inplace=True)
          (conv1): Conv2d(1696, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (norm2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (relu2): ReLU(inplace=True)
          (conv2): Conv2d(128, 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
        )
        (denselayer47): _DenseLayer(
          (norm1): BatchNorm2d(1728, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (relu1): ReLU(inplace=True)
          (conv1): Conv2d(1728, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (norm2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (relu2): ReLU(inplace=True)
          (conv2): Conv2d(128, 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
        )
        (denselayer48): _DenseLayer(
          (norm1): BatchNorm2d(1760, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (relu1): ReLU(inplace=True)
          (conv1): Conv2d(1760, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (norm2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (relu2): ReLU(inplace=True)
          (conv2): Conv2d(128, 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
        )
      )
      (transition3): _Transition(
        (norm): BatchNorm2d(1792, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
        (relu): ReLU(inplace=True)
        (conv): Conv2d(1792, 896, kernel_size=(1, 1), stride=(1, 1), bias=False)
        (pool): AvgPool2d(kernel_size=2, stride=2, padding=0)
      )
      (denseblock4): _DenseBlock(
        (denselayer1): _DenseLayer(
          (norm1): BatchNorm2d(896, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (relu1): ReLU(inplace=True)
          (conv1): Conv2d(896, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (norm2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (relu2): ReLU(inplace=True)
          (conv2): Conv2d(128, 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
        )
        (denselayer2): _DenseLayer(
          (norm1): BatchNorm2d(928, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (relu1): ReLU(inplace=True)
          (conv1): Conv2d(928, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (norm2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (relu2): ReLU(inplace=True)
          (conv2): Conv2d(128, 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
        )
        (denselayer3): _DenseLayer(
          (norm1): BatchNorm2d(960, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (relu1): ReLU(inplace=True)
          (conv1): Conv2d(960, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (norm2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (relu2): ReLU(inplace=True)
          (conv2): Conv2d(128, 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
        )
        (denselayer4): _DenseLayer(
          (norm1): BatchNorm2d(992, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (relu1): ReLU(inplace=True)
          (conv1): Conv2d(992, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (norm2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (relu2): ReLU(inplace=True)
          (conv2): Conv2d(128, 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
        )
        (denselayer5): _DenseLayer(
          (norm1): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (relu1): ReLU(inplace=True)
          (conv1): Conv2d(1024, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (norm2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (relu2): ReLU(inplace=True)
          (conv2): Conv2d(128, 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
        )
        (denselayer6): _DenseLayer(
          (norm1): BatchNorm2d(1056, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (relu1): ReLU(inplace=True)
          (conv1): Conv2d(1056, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (norm2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (relu2): ReLU(inplace=True)
          (conv2): Conv2d(128, 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
        )
        (denselayer7): _DenseLayer(
          (norm1): BatchNorm2d(1088, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (relu1): ReLU(inplace=True)
          (conv1): Conv2d(1088, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (norm2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (relu2): ReLU(inplace=True)
          (conv2): Conv2d(128, 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
        )
        (denselayer8): _DenseLayer(
          (norm1): BatchNorm2d(1120, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (relu1): ReLU(inplace=True)
          (conv1): Conv2d(1120, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (norm2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (relu2): ReLU(inplace=True)
          (conv2): Conv2d(128, 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
        )
        (denselayer9): _DenseLayer(
          (norm1): BatchNorm2d(1152, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (relu1): ReLU(inplace=True)
          (conv1): Conv2d(1152, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (norm2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (relu2): ReLU(inplace=True)
          (conv2): Conv2d(128, 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
        )
        (denselayer10): _DenseLayer(
          (norm1): BatchNorm2d(1184, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (relu1): ReLU(inplace=True)
          (conv1): Conv2d(1184, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (norm2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (relu2): ReLU(inplace=True)
          (conv2): Conv2d(128, 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
        )
        (denselayer11): _DenseLayer(
          (norm1): BatchNorm2d(1216, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (relu1): ReLU(inplace=True)
          (conv1): Conv2d(1216, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (norm2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (relu2): ReLU(inplace=True)
          (conv2): Conv2d(128, 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
        )
        (denselayer12): _DenseLayer(
          (norm1): BatchNorm2d(1248, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (relu1): ReLU(inplace=True)
          (conv1): Conv2d(1248, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (norm2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (relu2): ReLU(inplace=True)
          (conv2): Conv2d(128, 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
        )
        (denselayer13): _DenseLayer(
          (norm1): BatchNorm2d(1280, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (relu1): ReLU(inplace=True)
          (conv1): Conv2d(1280, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (norm2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (relu2): ReLU(inplace=True)
          (conv2): Conv2d(128, 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
        )
        (denselayer14): _DenseLayer(
          (norm1): BatchNorm2d(1312, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (relu1): ReLU(inplace=True)
          (conv1): Conv2d(1312, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (norm2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (relu2): ReLU(inplace=True)
          (conv2): Conv2d(128, 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
        )
        (denselayer15): _DenseLayer(
          (norm1): BatchNorm2d(1344, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (relu1): ReLU(inplace=True)
          (conv1): Conv2d(1344, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (norm2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (relu2): ReLU(inplace=True)
          (conv2): Conv2d(128, 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
        )
        (denselayer16): _DenseLayer(
          (norm1): BatchNorm2d(1376, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (relu1): ReLU(inplace=True)
          (conv1): Conv2d(1376, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (norm2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (relu2): ReLU(inplace=True)
          (conv2): Conv2d(128, 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
        )
        (denselayer17): _DenseLayer(
          (norm1): BatchNorm2d(1408, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (relu1): ReLU(inplace=True)
          (conv1): Conv2d(1408, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (norm2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (relu2): ReLU(inplace=True)
          (conv2): Conv2d(128, 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
        )
        (denselayer18): _DenseLayer(
          (norm1): BatchNorm2d(1440, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (relu1): ReLU(inplace=True)
          (conv1): Conv2d(1440, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (norm2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (relu2): ReLU(inplace=True)
          (conv2): Conv2d(128, 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
        )
        (denselayer19): _DenseLayer(
          (norm1): BatchNorm2d(1472, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (relu1): ReLU(inplace=True)
          (conv1): Conv2d(1472, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (norm2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (relu2): ReLU(inplace=True)
          (conv2): Conv2d(128, 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
        )
        (denselayer20): _DenseLayer(
          (norm1): BatchNorm2d(1504, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (relu1): ReLU(inplace=True)
          (conv1): Conv2d(1504, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (norm2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (relu2): ReLU(inplace=True)
          (conv2): Conv2d(128, 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
        )
        (denselayer21): _DenseLayer(
          (norm1): BatchNorm2d(1536, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (relu1): ReLU(inplace=True)
          (conv1): Conv2d(1536, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (norm2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (relu2): ReLU(inplace=True)
          (conv2): Conv2d(128, 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
        )
        (denselayer22): _DenseLayer(
          (norm1): BatchNorm2d(1568, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (relu1): ReLU(inplace=True)
          (conv1): Conv2d(1568, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (norm2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (relu2): ReLU(inplace=True)
          (conv2): Conv2d(128, 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
        )
        (denselayer23): _DenseLayer(
          (norm1): BatchNorm2d(1600, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (relu1): ReLU(inplace=True)
          (conv1): Conv2d(1600, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (norm2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (relu2): ReLU(inplace=True)
          (conv2): Conv2d(128, 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
        )
        (denselayer24): _DenseLayer(
          (norm1): BatchNorm2d(1632, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (relu1): ReLU(inplace=True)
          (conv1): Conv2d(1632, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (norm2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (relu2): ReLU(inplace=True)
          (conv2): Conv2d(128, 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
        )
        (denselayer25): _DenseLayer(
          (norm1): BatchNorm2d(1664, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (relu1): ReLU(inplace=True)
          (conv1): Conv2d(1664, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (norm2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (relu2): ReLU(inplace=True)
          (conv2): Conv2d(128, 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
        )
        (denselayer26): _DenseLayer(
          (norm1): BatchNorm2d(1696, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (relu1): ReLU(inplace=True)
          (conv1): Conv2d(1696, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (norm2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (relu2): ReLU(inplace=True)
          (conv2): Conv2d(128, 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
        )
        (denselayer27): _DenseLayer(
          (norm1): BatchNorm2d(1728, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (relu1): ReLU(inplace=True)
          (conv1): Conv2d(1728, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (norm2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (relu2): ReLU(inplace=True)
          (conv2): Conv2d(128, 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
        )
        (denselayer28): _DenseLayer(
          (norm1): BatchNorm2d(1760, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (relu1): ReLU(inplace=True)
          (conv1): Conv2d(1760, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (norm2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (relu2): ReLU(inplace=True)
          (conv2): Conv2d(128, 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
        )
        (denselayer29): _DenseLayer(
          (norm1): BatchNorm2d(1792, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (relu1): ReLU(inplace=True)
          (conv1): Conv2d(1792, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (norm2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (relu2): ReLU(inplace=True)
          (conv2): Conv2d(128, 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
        )
        (denselayer30): _DenseLayer(
          (norm1): BatchNorm2d(1824, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (relu1): ReLU(inplace=True)
          (conv1): Conv2d(1824, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (norm2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (relu2): ReLU(inplace=True)
          (conv2): Conv2d(128, 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
        )
        (denselayer31): _DenseLayer(
          (norm1): BatchNorm2d(1856, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (relu1): ReLU(inplace=True)
          (conv1): Conv2d(1856, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (norm2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (relu2): ReLU(inplace=True)
          (conv2): Conv2d(128, 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
        )
        (denselayer32): _DenseLayer(
          (norm1): BatchNorm2d(1888, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (relu1): ReLU(inplace=True)
          (conv1): Conv2d(1888, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (norm2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (relu2): ReLU(inplace=True)
          (conv2): Conv2d(128, 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
        )
      )
      (norm5): BatchNorm2d(1920, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
    )
  )
  (1): Sequential(
    (0): AdaptiveConcatPool2d(
      (ap): AdaptiveAvgPool2d(output_size=1)
      (mp): AdaptiveMaxPool2d(output_size=1)
    )
    (1): Flatten()
    (2): BatchNorm1d(3840, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
    (3): Dropout(p=0.25, inplace=False)
    (4): Linear(in_features=3840, out_features=512, bias=True)
    (5): ReLU(inplace=True)
    (6): BatchNorm1d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
    (7): Dropout(p=0.5, inplace=False)
    (8): Linear(in_features=512, out_features=3, bias=True)
  )
), opt_func=functools.partial(<class 'torch.optim.adam.Adam'>, betas=(0.9, 0.99)), loss_func=FlattenedLoss of CrossEntropyLoss(), metrics=[<function accuracy at 0x7fde9d3c6158>], true_wd=True, bn_wd=True, wd=0.01, train_bn=True, path=PosixPath('.'), model_dir='models', callback_fns=[functools.partial(<class 'fastai.basic_train.Recorder'>, add_time=True, silent=False), functools.partial(<class 'fastai.callbacks.mixup.MixUpCallback'>, alpha=0.4, stack_x=False, stack_y=True)], callbacks=[], layer_groups=[Sequential(
  (0): Conv2d(3, 64, kernel_size=(7, 7), stride=(2, 2), padding=(3, 3), bias=False)
  (1): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
  (2): ReLU(inplace=True)
  (3): MaxPool2d(kernel_size=3, stride=2, padding=1, dilation=1, ceil_mode=False)
  (4): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
  (5): ReLU(inplace=True)
  (6): Conv2d(64, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)
  (7): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
  (8): ReLU(inplace=True)
  (9): Conv2d(128, 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
  (10): BatchNorm2d(96, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
  (11): ReLU(inplace=True)
  (12): Conv2d(96, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)
  (13): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
  (14): ReLU(inplace=True)
  (15): Conv2d(128, 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
  (16): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
  (17): ReLU(inplace=True)
  (18): Conv2d(128, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)
  (19): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
  (20): ReLU(inplace=True)
  (21): Conv2d(128, 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
  (22): BatchNorm2d(160, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
  (23): ReLU(inplace=True)
  (24): Conv2d(160, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)
  (25): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
  (26): ReLU(inplace=True)
  (27): Conv2d(128, 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
  (28): BatchNorm2d(192, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
  (29): ReLU(inplace=True)
  (30): Conv2d(192, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)
  (31): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
  (32): ReLU(inplace=True)
  (33): Conv2d(128, 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
  (34): BatchNorm2d(224, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
  (35): ReLU(inplace=True)
  (36): Conv2d(224, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)
  (37): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
  (38): ReLU(inplace=True)
  (39): Conv2d(128, 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
  (40): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
  (41): ReLU(inplace=True)
  (42): Conv2d(256, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)
  (43): AvgPool2d(kernel_size=2, stride=2, padding=0)
  (44): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
  (45): ReLU(inplace=True)
  (46): Conv2d(128, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)
  (47): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
  (48): ReLU(inplace=True)
  (49): Conv2d(128, 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
  (50): BatchNorm2d(160, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
  (51): ReLU(inplace=True)
  (52): Conv2d(160, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)
  (53): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
  (54): ReLU(inplace=True)
  (55): Conv2d(128, 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
  (56): BatchNorm2d(192, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
  (57): ReLU(inplace=True)
  (58): Conv2d(192, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)
  (59): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
  (60): ReLU(inplace=True)
  (61): Conv2d(128, 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
  (62): BatchNorm2d(224, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
  (63): ReLU(inplace=True)
  (64): Conv2d(224, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)
  (65): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
  (66): ReLU(inplace=True)
  (67): Conv2d(128, 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
  (68): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
  (69): ReLU(inplace=True)
  (70): Conv2d(256, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)
  (71): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
  (72): ReLU(inplace=True)
  (73): Conv2d(128, 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
  (74): BatchNorm2d(288, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
  (75): ReLU(inplace=True)
  (76): Conv2d(288, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)
  (77): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
  (78): ReLU(inplace=True)
  (79): Conv2d(128, 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
  (80): BatchNorm2d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
  (81): ReLU(inplace=True)
  (82): Conv2d(320, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)
  (83): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
  (84): ReLU(inplace=True)
  (85): Conv2d(128, 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
  (86): BatchNorm2d(352, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
  (87): ReLU(inplace=True)
  (88): Conv2d(352, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)
  (89): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
  (90): ReLU(inplace=True)
  (91): Conv2d(128, 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
  (92): BatchNorm2d(384, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
  (93): ReLU(inplace=True)
  (94): Conv2d(384, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)
  (95): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
  (96): ReLU(inplace=True)
  (97): Conv2d(128, 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
  (98): BatchNorm2d(416, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
  (99): ReLU(inplace=True)
  (100): Conv2d(416, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)
  (101): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
  (102): ReLU(inplace=True)
  (103): Conv2d(128, 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
  (104): BatchNorm2d(448, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
  (105): ReLU(inplace=True)
  (106): Conv2d(448, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)
  (107): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
  (108): ReLU(inplace=True)
  (109): Conv2d(128, 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
  (110): BatchNorm2d(480, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
  (111): ReLU(inplace=True)
  (112): Conv2d(480, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)
  (113): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
  (114): ReLU(inplace=True)
  (115): Conv2d(128, 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
), Sequential(
  (0): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
  (1): ReLU(inplace=True)
  (2): Conv2d(512, 256, kernel_size=(1, 1), stride=(1, 1), bias=False)
  (3): AvgPool2d(kernel_size=2, stride=2, padding=0)
  (4): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
  (5): ReLU(inplace=True)
  (6): Conv2d(256, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)
  (7): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
  (8): ReLU(inplace=True)
  (9): Conv2d(128, 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
  (10): BatchNorm2d(288, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
  (11): ReLU(inplace=True)
  (12): Conv2d(288, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)
  (13): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
  (14): ReLU(inplace=True)
  (15): Conv2d(128, 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
  (16): BatchNorm2d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
  (17): ReLU(inplace=True)
  (18): Conv2d(320, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)
  (19): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
  (20): ReLU(inplace=True)
  (21): Conv2d(128, 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
  (22): BatchNorm2d(352, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
  (23): ReLU(inplace=True)
  (24): Conv2d(352, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)
  (25): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
  (26): ReLU(inplace=True)
  (27): Conv2d(128, 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
  (28): BatchNorm2d(384, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
  (29): ReLU(inplace=True)
  (30): Conv2d(384, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)
  (31): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
  (32): ReLU(inplace=True)
  (33): Conv2d(128, 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
  (34): BatchNorm2d(416, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
  (35): ReLU(inplace=True)
  (36): Conv2d(416, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)
  (37): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
  (38): ReLU(inplace=True)
  (39): Conv2d(128, 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
  (40): BatchNorm2d(448, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
  (41): ReLU(inplace=True)
  (42): Conv2d(448, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)
  (43): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
  (44): ReLU(inplace=True)
  (45): Conv2d(128, 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
  (46): BatchNorm2d(480, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
  (47): ReLU(inplace=True)
  (48): Conv2d(480, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)
  (49): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
  (50): ReLU(inplace=True)
  (51): Conv2d(128, 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
  (52): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
  (53): ReLU(inplace=True)
  (54): Conv2d(512, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)
  (55): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
  (56): ReLU(inplace=True)
  (57): Conv2d(128, 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
  (58): BatchNorm2d(544, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
  (59): ReLU(inplace=True)
  (60): Conv2d(544, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)
  (61): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
  (62): ReLU(inplace=True)
  (63): Conv2d(128, 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
  (64): BatchNorm2d(576, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
  (65): ReLU(inplace=True)
  (66): Conv2d(576, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)
  (67): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
  (68): ReLU(inplace=True)
  (69): Conv2d(128, 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
  (70): BatchNorm2d(608, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
  (71): ReLU(inplace=True)
  (72): Conv2d(608, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)
  (73): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
  (74): ReLU(inplace=True)
  (75): Conv2d(128, 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
  (76): BatchNorm2d(640, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
  (77): ReLU(inplace=True)
  (78): Conv2d(640, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)
  (79): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
  (80): ReLU(inplace=True)
  (81): Conv2d(128, 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
  (82): BatchNorm2d(672, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
  (83): ReLU(inplace=True)
  (84): Conv2d(672, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)
  (85): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
  (86): ReLU(inplace=True)
  (87): Conv2d(128, 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
  (88): BatchNorm2d(704, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
  (89): ReLU(inplace=True)
  (90): Conv2d(704, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)
  (91): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
  (92): ReLU(inplace=True)
  (93): Conv2d(128, 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
  (94): BatchNorm2d(736, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
  (95): ReLU(inplace=True)
  (96): Conv2d(736, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)
  (97): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
  (98): ReLU(inplace=True)
  (99): Conv2d(128, 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
  (100): BatchNorm2d(768, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
  (101): ReLU(inplace=True)
  (102): Conv2d(768, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)
  (103): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
  (104): ReLU(inplace=True)
  (105): Conv2d(128, 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
  (106): BatchNorm2d(800, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
  (107): ReLU(inplace=True)
  (108): Conv2d(800, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)
  (109): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
  (110): ReLU(inplace=True)
  (111): Conv2d(128, 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
  (112): BatchNorm2d(832, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
  (113): ReLU(inplace=True)
  (114): Conv2d(832, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)
  (115): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
  (116): ReLU(inplace=True)
  (117): Conv2d(128, 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
  (118): BatchNorm2d(864, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
  (119): ReLU(inplace=True)
  (120): Conv2d(864, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)
  (121): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
  (122): ReLU(inplace=True)
  (123): Conv2d(128, 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
  (124): BatchNorm2d(896, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
  (125): ReLU(inplace=True)
  (126): Conv2d(896, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)
  (127): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
  (128): ReLU(inplace=True)
  (129): Conv2d(128, 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
  (130): BatchNorm2d(928, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
  (131): ReLU(inplace=True)
  (132): Conv2d(928, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)
  (133): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
  (134): ReLU(inplace=True)
  (135): Conv2d(128, 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
  (136): BatchNorm2d(960, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
  (137): ReLU(inplace=True)
  (138): Conv2d(960, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)
  (139): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
  (140): ReLU(inplace=True)
  (141): Conv2d(128, 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
  (142): BatchNorm2d(992, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
  (143): ReLU(inplace=True)
  (144): Conv2d(992, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)
  (145): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
  (146): ReLU(inplace=True)
  (147): Conv2d(128, 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
  (148): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
  (149): ReLU(inplace=True)
  (150): Conv2d(1024, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)
  (151): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
  (152): ReLU(inplace=True)
  (153): Conv2d(128, 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
  (154): BatchNorm2d(1056, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
  (155): ReLU(inplace=True)
  (156): Conv2d(1056, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)
  (157): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
  (158): ReLU(inplace=True)
  (159): Conv2d(128, 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
  (160): BatchNorm2d(1088, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
  (161): ReLU(inplace=True)
  (162): Conv2d(1088, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)
  (163): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
  (164): ReLU(inplace=True)
  (165): Conv2d(128, 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
  (166): BatchNorm2d(1120, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
  (167): ReLU(inplace=True)
  (168): Conv2d(1120, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)
  (169): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
  (170): ReLU(inplace=True)
  (171): Conv2d(128, 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
  (172): BatchNorm2d(1152, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
  (173): ReLU(inplace=True)
  (174): Conv2d(1152, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)
  (175): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
  (176): ReLU(inplace=True)
  (177): Conv2d(128, 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
  (178): BatchNorm2d(1184, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
  (179): ReLU(inplace=True)
  (180): Conv2d(1184, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)
  (181): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
  (182): ReLU(inplace=True)
  (183): Conv2d(128, 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
  (184): BatchNorm2d(1216, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
  (185): ReLU(inplace=True)
  (186): Conv2d(1216, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)
  (187): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
  (188): ReLU(inplace=True)
  (189): Conv2d(128, 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
  (190): BatchNorm2d(1248, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
  (191): ReLU(inplace=True)
  (192): Conv2d(1248, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)
  (193): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
  (194): ReLU(inplace=True)
  (195): Conv2d(128, 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
  (196): BatchNorm2d(1280, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
  (197): ReLU(inplace=True)
  (198): Conv2d(1280, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)
  (199): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
  (200): ReLU(inplace=True)
  (201): Conv2d(128, 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
  (202): BatchNorm2d(1312, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
  (203): ReLU(inplace=True)
  (204): Conv2d(1312, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)
  (205): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
  (206): ReLU(inplace=True)
  (207): Conv2d(128, 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
  (208): BatchNorm2d(1344, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
  (209): ReLU(inplace=True)
  (210): Conv2d(1344, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)
  (211): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
  (212): ReLU(inplace=True)
  (213): Conv2d(128, 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
  (214): BatchNorm2d(1376, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
  (215): ReLU(inplace=True)
  (216): Conv2d(1376, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)
  (217): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
  (218): ReLU(inplace=True)
  (219): Conv2d(128, 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
  (220): BatchNorm2d(1408, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
  (221): ReLU(inplace=True)
  (222): Conv2d(1408, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)
  (223): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
  (224): ReLU(inplace=True)
  (225): Conv2d(128, 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
  (226): BatchNorm2d(1440, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
  (227): ReLU(inplace=True)
  (228): Conv2d(1440, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)
  (229): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
  (230): ReLU(inplace=True)
  (231): Conv2d(128, 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
  (232): BatchNorm2d(1472, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
  (233): ReLU(inplace=True)
  (234): Conv2d(1472, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)
  (235): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
  (236): ReLU(inplace=True)
  (237): Conv2d(128, 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
  (238): BatchNorm2d(1504, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
  (239): ReLU(inplace=True)
  (240): Conv2d(1504, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)
  (241): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
  (242): ReLU(inplace=True)
  (243): Conv2d(128, 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
  (244): BatchNorm2d(1536, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
  (245): ReLU(inplace=True)
  (246): Conv2d(1536, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)
  (247): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
  (248): ReLU(inplace=True)
  (249): Conv2d(128, 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
  (250): BatchNorm2d(1568, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
  (251): ReLU(inplace=True)
  (252): Conv2d(1568, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)
  (253): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
  (254): ReLU(inplace=True)
  (255): Conv2d(128, 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
  (256): BatchNorm2d(1600, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
  (257): ReLU(inplace=True)
  (258): Conv2d(1600, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)
  (259): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
  (260): ReLU(inplace=True)
  (261): Conv2d(128, 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
  (262): BatchNorm2d(1632, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
  (263): ReLU(inplace=True)
  (264): Conv2d(1632, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)
  (265): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
  (266): ReLU(inplace=True)
  (267): Conv2d(128, 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
  (268): BatchNorm2d(1664, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
  (269): ReLU(inplace=True)
  (270): Conv2d(1664, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)
  (271): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
  (272): ReLU(inplace=True)
  (273): Conv2d(128, 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
  (274): BatchNorm2d(1696, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
  (275): ReLU(inplace=True)
  (276): Conv2d(1696, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)
  (277): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
  (278): ReLU(inplace=True)
  (279): Conv2d(128, 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
  (280): BatchNorm2d(1728, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
  (281): ReLU(inplace=True)
  (282): Conv2d(1728, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)
  (283): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
  (284): ReLU(inplace=True)
  (285): Conv2d(128, 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
  (286): BatchNorm2d(1760, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
  (287): ReLU(inplace=True)
  (288): Conv2d(1760, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)
  (289): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
  (290): ReLU(inplace=True)
  (291): Conv2d(128, 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
  (292): BatchNorm2d(1792, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
  (293): ReLU(inplace=True)
  (294): Conv2d(1792, 896, kernel_size=(1, 1), stride=(1, 1), bias=False)
  (295): AvgPool2d(kernel_size=2, stride=2, padding=0)
  (296): BatchNorm2d(896, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
  (297): ReLU(inplace=True)
  (298): Conv2d(896, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)
  (299): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
  (300): ReLU(inplace=True)
  (301): Conv2d(128, 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
  (302): BatchNorm2d(928, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
  (303): ReLU(inplace=True)
  (304): Conv2d(928, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)
  (305): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
  (306): ReLU(inplace=True)
  (307): Conv2d(128, 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
  (308): BatchNorm2d(960, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
  (309): ReLU(inplace=True)
  (310): Conv2d(960, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)
  (311): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
  (312): ReLU(inplace=True)
  (313): Conv2d(128, 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
  (314): BatchNorm2d(992, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
  (315): ReLU(inplace=True)
  (316): Conv2d(992, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)
  (317): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
  (318): ReLU(inplace=True)
  (319): Conv2d(128, 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
  (320): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
  (321): ReLU(inplace=True)
  (322): Conv2d(1024, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)
  (323): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
  (324): ReLU(inplace=True)
  (325): Conv2d(128, 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
  (326): BatchNorm2d(1056, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
  (327): ReLU(inplace=True)
  (328): Conv2d(1056, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)
  (329): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
  (330): ReLU(inplace=True)
  (331): Conv2d(128, 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
  (332): BatchNorm2d(1088, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
  (333): ReLU(inplace=True)
  (334): Conv2d(1088, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)
  (335): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
  (336): ReLU(inplace=True)
  (337): Conv2d(128, 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
  (338): BatchNorm2d(1120, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
  (339): ReLU(inplace=True)
  (340): Conv2d(1120, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)
  (341): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
  (342): ReLU(inplace=True)
  (343): Conv2d(128, 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
  (344): BatchNorm2d(1152, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
  (345): ReLU(inplace=True)
  (346): Conv2d(1152, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)
  (347): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
  (348): ReLU(inplace=True)
  (349): Conv2d(128, 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
  (350): BatchNorm2d(1184, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
  (351): ReLU(inplace=True)
  (352): Conv2d(1184, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)
  (353): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
  (354): ReLU(inplace=True)
  (355): Conv2d(128, 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
  (356): BatchNorm2d(1216, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
  (357): ReLU(inplace=True)
  (358): Conv2d(1216, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)
  (359): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
  (360): ReLU(inplace=True)
  (361): Conv2d(128, 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
  (362): BatchNorm2d(1248, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
  (363): ReLU(inplace=True)
  (364): Conv2d(1248, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)
  (365): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
  (366): ReLU(inplace=True)
  (367): Conv2d(128, 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
  (368): BatchNorm2d(1280, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
  (369): ReLU(inplace=True)
  (370): Conv2d(1280, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)
  (371): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
  (372): ReLU(inplace=True)
  (373): Conv2d(128, 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
  (374): BatchNorm2d(1312, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
  (375): ReLU(inplace=True)
  (376): Conv2d(1312, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)
  (377): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
  (378): ReLU(inplace=True)
  (379): Conv2d(128, 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
  (380): BatchNorm2d(1344, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
  (381): ReLU(inplace=True)
  (382): Conv2d(1344, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)
  (383): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
  (384): ReLU(inplace=True)
  (385): Conv2d(128, 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
  (386): BatchNorm2d(1376, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
  (387): ReLU(inplace=True)
  (388): Conv2d(1376, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)
  (389): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
  (390): ReLU(inplace=True)
  (391): Conv2d(128, 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
  (392): BatchNorm2d(1408, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
  (393): ReLU(inplace=True)
  (394): Conv2d(1408, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)
  (395): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
  (396): ReLU(inplace=True)
  (397): Conv2d(128, 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
  (398): BatchNorm2d(1440, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
  (399): ReLU(inplace=True)
  (400): Conv2d(1440, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)
  (401): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
  (402): ReLU(inplace=True)
  (403): Conv2d(128, 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
  (404): BatchNorm2d(1472, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
  (405): ReLU(inplace=True)
  (406): Conv2d(1472, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)
  (407): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
  (408): ReLU(inplace=True)
  (409): Conv2d(128, 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
  (410): BatchNorm2d(1504, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
  (411): ReLU(inplace=True)
  (412): Conv2d(1504, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)
  (413): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
  (414): ReLU(inplace=True)
  (415): Conv2d(128, 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
  (416): BatchNorm2d(1536, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
  (417): ReLU(inplace=True)
  (418): Conv2d(1536, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)
  (419): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
  (420): ReLU(inplace=True)
  (421): Conv2d(128, 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
  (422): BatchNorm2d(1568, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
  (423): ReLU(inplace=True)
  (424): Conv2d(1568, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)
  (425): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
  (426): ReLU(inplace=True)
  (427): Conv2d(128, 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
  (428): BatchNorm2d(1600, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
  (429): ReLU(inplace=True)
  (430): Conv2d(1600, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)
  (431): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
  (432): ReLU(inplace=True)
  (433): Conv2d(128, 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
  (434): BatchNorm2d(1632, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
  (435): ReLU(inplace=True)
  (436): Conv2d(1632, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)
  (437): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
  (438): ReLU(inplace=True)
  (439): Conv2d(128, 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
  (440): BatchNorm2d(1664, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
  (441): ReLU(inplace=True)
  (442): Conv2d(1664, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)
  (443): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
  (444): ReLU(inplace=True)
  (445): Conv2d(128, 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
  (446): BatchNorm2d(1696, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
  (447): ReLU(inplace=True)
  (448): Conv2d(1696, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)
  (449): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
  (450): ReLU(inplace=True)
  (451): Conv2d(128, 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
  (452): BatchNorm2d(1728, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
  (453): ReLU(inplace=True)
  (454): Conv2d(1728, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)
  (455): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
  (456): ReLU(inplace=True)
  (457): Conv2d(128, 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
  (458): BatchNorm2d(1760, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
  (459): ReLU(inplace=True)
  (460): Conv2d(1760, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)
  (461): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
  (462): ReLU(inplace=True)
  (463): Conv2d(128, 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
  (464): BatchNorm2d(1792, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
  (465): ReLU(inplace=True)
  (466): Conv2d(1792, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)
  (467): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
  (468): ReLU(inplace=True)
  (469): Conv2d(128, 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
  (470): BatchNorm2d(1824, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
  (471): ReLU(inplace=True)
  (472): Conv2d(1824, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)
  (473): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
  (474): ReLU(inplace=True)
  (475): Conv2d(128, 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
  (476): BatchNorm2d(1856, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
  (477): ReLU(inplace=True)
  (478): Conv2d(1856, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)
  (479): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
  (480): ReLU(inplace=True)
  (481): Conv2d(128, 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
  (482): BatchNorm2d(1888, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
  (483): ReLU(inplace=True)
  (484): Conv2d(1888, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)
  (485): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
  (486): ReLU(inplace=True)
  (487): Conv2d(128, 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
  (488): BatchNorm2d(1920, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
), Sequential(
  (0): AdaptiveAvgPool2d(output_size=1)
  (1): AdaptiveMaxPool2d(output_size=1)
  (2): Flatten()
  (3): BatchNorm1d(3840, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
  (4): Dropout(p=0.25, inplace=False)
  (5): Linear(in_features=3840, out_features=512, bias=True)
  (6): ReLU(inplace=True)
  (7): BatchNorm1d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
  (8): Dropout(p=0.5, inplace=False)
  (9): Linear(in_features=512, out_features=3, bias=True)
)], add_time=True, silent=False)














    

    
    







    


learner.data



    








    

    
    







    


test = ImageList.from_df(test_df,path='/')
learner.data= (ImageList.from_df(df, path='/')
              .split_by_idx(val_idx)
              .label_from_df(cols='disease_class')
              .add_test(test)
              .transform(get_transforms(flip_vert=True), size=512)
              .databunch(path='.', bs=8)
              .normalize(imagenet_stats)
              )
learner.data



    



















ImageDataBunch;

Train: LabelList (788 items)
x: ImageList
Image (3, 512, 512),Image (3, 512, 512),Image (3, 512, 512),Image (3, 512, 512),Image (3, 512, 512)
y: CategoryList
healthy_wheat,healthy_wheat,healthy_wheat,healthy_wheat,healthy_wheat
Path: /;

Valid: LabelList (88 items)
x: ImageList
Image (3, 512, 512),Image (3, 512, 512),Image (3, 512, 512),Image (3, 512, 512),Image (3, 512, 512)
y: CategoryList
stem_rust,stem_rust,leaf_rust,leaf_rust,leaf_rust
Path: /;

Test: LabelList (610 items)
x: ImageList
Image (3, 512, 512),Image (3, 512, 512),Image (3, 512, 512),Image (3, 512, 512),Image (3, 512, 512)
y: EmptyLabelList
,,,,
Path: /














    

    
    







    


learner.freeze()
learner.fit_one_cycle(3, slice(1e-5/400,1e-3/4))



    



















  
    

      epoch
      train_loss
      valid_loss
      accuracy
      time
    

  
  
    

      0
      0.825689
      0.545968
      0.840909
      03:36
    

    

      1
      0.819339
      0.531960
      0.863636
      03:42
    

    

      2
      0.825667
      0.476919
      0.863636
      03:39
    

  















    

    
    







    


learner.unfreeze()
learner.fit_one_cycle(10, slice(1e-5/400,1e-3/4))



    



















  
    

      epoch
      train_loss
      valid_loss
      accuracy
      time
    

  
  
    

      0
      0.804510
      0.507069
      0.863636
      03:39
    

    

      1
      0.772364
      0.505754
      0.852273
      03:37
    

    

      2
      0.821464
      0.506186
      0.840909
      03:39
    

    

      3
      0.777784
      0.496004
      0.852273
      03:35
    

    

      4
      0.810471
      0.501964
      0.840909
      03:39
    

    

      5
      0.714839
      0.493431
      0.840909
      03:38
    

    

      6
      0.726852
      0.483276
      0.840909
      03:38
    

    

      7
      0.767355
      0.488357
      0.840909
      03:39
    

    

      8
      0.715276
      0.471205
      0.852273
      03:40
    

    

      9
      0.736499
      0.470084
      0.875000
      03:39
    

  















    

    
    







    


preds,y = learner.TTA()



    












































    

    
    







    


print('The accuracy using TTA is:',accuracy(preds, y).item())



    

















The accuracy using TTA is: 0.875














    

    
    







    


#Prediction of the Test Set
preds_test, y_test = learner.TTA(ds_type=DatasetType.Test)



    












































    

    
    







    


submission = pd.DataFrame({'ID': [s.split('.')[0] for s in os.listdir('/content/drive/My Drive/ICLR/ICLR/test/test')  if s.split('.')[1] != 'jfif']})
for i, c in enumerate(learner.data.classes):
  submission[c] = preds_test[:,i]
submission = submission[['ID', 'leaf_rust', 'stem_rust', 'healthy_wheat']] # Get same order as sample sub
submission.head()



    






















  
    

      
      ID
      leaf_rust
      stem_rust
      healthy_wheat
    

  
  
    

      0
      008FWT
      0.688015
      0.267412
      0.044573
    

    

      1
      085IEC
      0.025334
      0.096655
      0.878011
    

    

      2
      00AQXY
      0.076655
      0.920853
      0.002492
    

    

      3
      01OJZX
      0.656178
      0.300063
      0.043759
    

    

      4
      0K9T9T
      0.390184
      0.596413
      0.013404
    

  

















    

    
    







    


submission.to_csv('/content/drive/My Drive/ICLR/submission.csv', index = False)



    








    

</div>