chainerでfine-tuningを行うとtrainer.run()でエラーが生じる

前提・実現したいこと

犬と猫の画像をCNNで分類するシステムを作っています。
環境はgoogle colaboratoryで、フレームワークはchainerを使っています。
また、画像はgoogle driveから読み込み、ネットワークはvggを使っています。
しかし、trainer.run()を実行すると下記のようなエラーが出てしまいます。

発生している問題・エラーメッセージ

ValueError                                Traceback (most recent call last)
<ipython-input-23-041e2033e90a> in <module>()
----> 1 trainer.run()

8 frames
/usr/local/lib/python3.6/dist-packages/chainer/dataset/convert.py in _concat_arrays(arrays, padding)
    252         with chainer.using_device(device):
    253             arr_concat = device.xp.concatenate(
--> 254                 [array[None] for array in arrays])
    255 
    256     return arr_concat

<__array_function__ internals> in concatenate(*args, **kwargs)

ValueError: all the input array dimensions for the concatenation axis must match exactly, but along dimension 1, the array at index 0 has size 3 and the array at index 8 has size 4

該当のソースコード

import glob
import chainer
import chainer.links as L
import chainer.functions as F
import numpy as np
import pandas as pd
from google.colab import drive
from PIL import Image
from sklearn.model_selection import train_test_split
from chainer.datasets import tuple_dataset
from chainer import training
from chainer.training import extensions


#google driveにマウント
drive.mount('/content/drive') 


#ファイルパス取得
path1 = glob.glob("/content/drive/My Drive/classification/dog*.jpg") 
path2 = glob.glob("/content/drive/My Drive/classification/cat*.jpg")


#ラベル配列作成
t1 = np.array( [] ) 
t2 = np.array( [] )

for i in range(len(path1)):
t1 = np.append( t1, 0 )

for i in range(len(path2)):
t2 = np.append( t2, 1 )

t1 = t1.astype('int32')
t2 = t2.astype('int32')


#画像前処理
imageData1 = []　
imageData2 = []

for i in range(len(path1)):
img = Image.open(path1[i])
img = img.resize((224, 224))
img = np.asarray(img)
img = img.transpose(2,1,0)
img = img.astype(np.float32)
img = img / 255
imageData1.append(img)

for i in range(len(path2)):
img = Image.open(path2[i])
img = img.resize((224, 224))
img = np.asarray(img)
img = img.transpose(2,1,0)
img = img.astype(np.float32)
img = img / 255
imageData2.append(img)


#データ整理
x_train1, x_test1, t_train1, t_test1 = train_test_split(imageData1, t1, test_size=0.3, random_state=0)
x_train1, x_test2, t_train2, t_test2 = train_test_split(imageData2, t2, test_size=0.3, random_state=0)

x_train = []
for i in range(len(x_train1)):
x_train.append(x_train1[i])
for i in range(len(x_train2)):
x_train.append(x_train2[i])

x_test = []
for i in range(len(x_test1)):
x_test.append(x_test1[i])
for i in range(len(x_test2)):
x_test.append(x_test2[i])

t_train = []
for i in range(len(t_train1)):
t_train.append(t_train1[i])
for i in range(len(t_train2)):
t_train.append(t_train2[i])

t_test = []
for i in range(len(t_test1)):
t_test.append(t_test1[i])
for i in range(len(t_test2)):
t_test.append(t_test2[i])

train = tuple_dataset.TupleDataset(x_train, t_train)
test = tuple_dataset.TupleDataset(x_test, t_test)


#ネットワーク
class VGG(chainer.Chain):
  def __init__(self, class_labels=2):
      super(VGG, self).__init__()

      with self.init_scope():
          self.base = L.VGG16Layers()
          self.fc6 = L.Linear(512 * 7 * 7, 4096)
          self.fc7 = L.Linear(4096, 4096)
          self.fc8 = L.Linear(4096, class_labels)

  def __call__(self, x):
      h = self.base(x, layers=['pool5'])['pool5']
      h = F.dropout(F.relu(self.fc6(h)))
      h = F.dropout(F.relu(self.fc7(h)))
      return self.fc8(h)


#設定
batchsize = 32
train_iter = chainer.iterators.SerialIterator(train, batchsize)
test_iter = chainer.iterators.SerialIterator(test, batchsize,
                                           repeat=False, shuffle=False)

vgg = VGG()
model = L.Classifier(vgg)
vgg.base.disable_update()
#model.base.disable_update()

gpu = 0
if gpu >= 0:
  chainer.cuda.get_device(gpu).use() 
  model.to_gpu()  

opt = chainer.optimizers.Adam()
opt.setup(model)

epoch = 15

updater = training.StandardUpdater(train_iter, opt, device=gpu)

trainer = training.Trainer(updater, (epoch, 'epoch'), out='/cifar/result')
trainer.extend(extensions.Evaluator(test_iter, model, device=gpu))
trainer.extend(extensions.dump_graph('main/loss'))
trainer.extend(extensions.snapshot(), trigger=(epoch, 'epoch'))
trainer.extend(extensions.LogReport())
trainer.extend(extensions.PrintReport(
  ['epoch', 'main/loss', 'validation/main/loss',
   'main/accuracy', 'validation/main/accuracy']))
trainer.extend(extensions.ProgressBar())


%学習
trainer.run()