作为大学项目的一部分，我已经为Galaxy Zoo项目创建了CNN模型。我已经使用批处理生成器创建了训练和验证数据，因为数据集非常大。我通过CPU毫无问题地对其进行了培训。最近，我切换到Tensorflow-gpu并通过GPU对其进行了训练，而GPU的速度是指数级的。但是当模型到达第33个时期时，它会意外停止。我已经链接了代码。请帮忙。
我尝试了10个纪元，20个以此类推。它可以正常工作到第32个纪元，但在完成第33个纪元时便会停止。
#Code for Image processor and Batch generator:
def img_processor(path):
img = cv2.imread(path)
img = img[106:106*3,106:106*3,:]
img = cv2.resize(img,(106,106), interpolation = cv2.INTER_CUBIC)
return img
def Batch_generator(DIR):
for img in os.listdir(DIR):
im = img_processor(os.path.join(DIR,img))
ind = int(img[:-4])
y_train = df.loc[ind].values
X_train = np.array(im)
X_train = X_train.reshape(1,106,106,3)
y_train = y_train.reshape(1,37)
yield(X_train,y_train)
def Validation_generator(DIR):
for img in os.listdir(DIR):
im = img_processor(os.path.join(DIR,img))
ind = int(img[:-4])
y_valid = df.loc[ind].values
X_valid = np.array(im)
X_valid = X_valid.reshape(1,106,106,3)
y_valid = y_valid.reshape(1,37)
yield(X_valid,y_valid)
#Fitting the model:
batch_size = 32
steps_per_batch = int(len(os.listdir(train_path))/batch_size)
val_steps_per_batch = int(len(os.listdir(validate_path))/batch_size)
history = model.fit_generator(
Batch_generator(train_path),
samples_per_epoch=steps_per_batch,
epochs = 70,
validation_data=Validation_generator(validate_path),
nb_val_samples=val_steps_per_batch,
verbose = 1,
callbacks = [tensorboard]
)
Epoch 31/70
1924/1924 [==============================] - 173s 90ms/step - loss: 0.0220 - acc: 0.5541 - val_loss: 0.0196 - val_acc: 0.4969
Epoch 32/70
1924/1924 [==============================] - 173s 90ms/step - loss: 0.0217 - acc: 0.5660 - val_loss: 0.0187 - val_acc: 0.5825
Epoch 33/70
10/1924 [..............................] - ETA: 2:20 - loss: 0.0270 - acc: 0.3000
---------------------------------------------------------------------------
StopIteration Traceback (most recent call last)
<ipython-input-9-fccdb670230f> in <module>
6 nb_val_samples=val_steps_per_batch,
7 verbose = 1,
----> 8 callbacks = [tensorboard]
9 )
c:\users\shinigami shrek\appdata\local\programs\python\python36\lib\site-packages\keras\legacy\interfaces.py in wrapper(*args, **kwargs)
89 warnings.warn('Update your `' + object_name + '` call to the ' +
90 'Keras 2 API: ' + signature, stacklevel=2)
---> 91 return func(*args, **kwargs)
92 wrapper._original_function = func
93 return wrapper
c:\users\shinigami shrek\appdata\local\programs\python\python36\lib\site-packages\keras\engine\training.py in fit_generator(self, generator, steps_per_epoch, epochs, verbose, callbacks, validation_data, validation_steps, class_weight, max_queue_size, workers, use_multiprocessing, shuffle, initial_epoch)
1416 use_multiprocessing=use_multiprocessing,
1417 shuffle=shuffle,
-> 1418 initial_epoch=initial_epoch)
1419
1420 @interfaces.legacy_generator_methods_support
c:\users\shinigami shrek\appdata\local\programs\python\python36\lib\site-packages\keras\engine\training_generator.py in fit_generator(model, generator, steps_per_epoch, epochs, verbose, callbacks, validation_data, validation_steps, class_weight, max_queue_size, workers, use_multiprocessing, shuffle, initial_epoch)
179 batch_index = 0
180 while steps_done < steps_per_epoch:
--> 181 generator_output = next(output_generator)
182
183 if not hasattr(generator_output, '__len__'):
StopIteration: