tensorflow:Your input ran out of data; interrupting training. Make sure that your dataset or generator can generate at least steps

行走在理想边缘 2022-05-16 发布于四川

展开全文

tensorflow:Your input ran out of data; interrupting training. Make sure that your dataset or generator can generate at least steps_per_epoch * epochs batches

打赏

一、总结

一句话总结：

保证batch_size（图像增强中）*steps_per_epoch（fit中）小于等于训练样本数

train_generator = train_datagen.flow_from_directory(
    train_dir, # 目标目录
    target_size=(150, 150), # 将所有图像的大小调整为 150×150
    batch_size=20, # 因为使用了 binary_crossentropy 损失，所以需要用二进制标签
    class_mode='binary')

history = model.fit(
    train_generator,
    steps_per_epoch=100,
    epochs=150,
    validation_data=validation_generator,
    validation_steps=50)

# case 1
# 如果上面train_generator的batch_size是32，如果这里steps_per_epoch=100，那么会报错
"""
tensorflow:Your input ran out of data; interrupting training.
Make sure that your dataset or generator can generate at least `steps_per_epoch * epochs` batches (in this case, 50 batches).
You may need to use the repeat() function when building your dataset.
"""
# 因为train样本数是2000（猫1000，狗1000），小于100*32
# case 2
# 如果上面train_generator的batch_size是20，如果这里steps_per_epoch=100，那么不会报错
# 因为大小刚好
# case 3
# 如果上面train_generator的batch_size是32，如果这里steps_per_epoch=int(1000/32)，
# 那么不会报错，但是会有警告，因为也是不整除
# 不会报错因为int(1000/32)*32 < 2000
# case 4
# 如果上面train_generator的batch_size是40，如果这里steps_per_epoch=100，照样报错
# 因为40*100>2000

回到顶部

二、tensorflow:Your input ran out of data; interrupting training. Make sure that your dataset or generator can generate at least steps_per_epoch * epochs batches

转自或参考：https:///questions/60509425/how-to-use-repeat-function-when-building-data-in-keras

1、报错

WARNING:tensorflow:Your input ran out of data;
interrupting training. Make sure that your dataset or generator can generate at least
steps_per_epoch * epochs batches (in this case, 5000 batches).
You may need to use the repeat() function when building your dataset.

2、现象

I am training a binary classifier on a dataset of cats and dogs:
Total Dataset: 10000 images
Training Dataset: 8000 images
Validation/Test Dataset: 2000 images

The Jupyter notebook code:

# Part 2 - Fitting the CNN to the images
train_datagen = ImageDataGenerator(rescale = 1./255,
                                   shear_range = 0.2,
                                   zoom_range = 0.2,
                                   horizontal_flip = True)

test_datagen = ImageDataGenerator(rescale = 1./255)

training_set = train_datagen.flow_from_directory('dataset/training_set',
                                                 target_size = (64, 64),
                                                 batch_size = 32,
                                                 class_mode = 'binary')

test_set = test_datagen.flow_from_directory('dataset/test_set',
                                            target_size = (64, 64),
                                            batch_size = 32,
                                            class_mode = 'binary')

history = model.fit_generator(training_set,
                              steps_per_epoch=8000,
                              epochs=25,
                              validation_data=test_set,
                              validation_steps=2000)

I trained it on a CPU without a problem but when I run on GPU it throws me this error:

Found 8000 images belonging to 2 classes.
Found 2000 images belonging to 2 classes.
WARNING:tensorflow:From <ipython-input-8-140743827a71>:23: Model.fit_generator (from tensorflow.python.keras.engine.training) is deprecated and will be removed in a future version.
Instructions for updating:
Please use Model.fit, which supports generators.
WARNING:tensorflow:sample_weight modes were coerced from
  ...
    to  
  ['...']
WARNING:tensorflow:sample_weight modes were coerced from
  ...
    to  
  ['...']
Train for 8000 steps, validate for 2000 steps
Epoch 1/25
 250/8000 [..............................] - ETA: 21:50 - loss: 7.6246 - accuracy: 0.5000
WARNING:tensorflow:Your input ran out of data; interrupting training. Make sure that your dataset or generator can generate at least `steps_per_epoch * epochs` batches (in this case, 200000 batches). You may need to use the repeat() function when building your dataset.
 250/8000 [..............................] - ETA: 21:52 - loss: 7.6246 - accuracy: 0.5000

I would like to know how to use the repeat() function in keras using Tensorflow 2.0?

3、解决

Your problem stems from the fact that the parameters steps_per_epoch and validation_steps need to be equal to the total number of data points divided to the batch_size.

Your code would work in Keras 1.X, prior to August 2017.

Change your model.fit function to:

history = model.fit_generator(training_set,
                              steps_per_epoch=int(8000/batch_size),
                              epochs=25,
                              validation_data=test_set,
                              validation_steps=int(2000/batch_size))

As of TensorFlow2.1, fit_generator is being deprecated. You can use .fit() method also on generators.

TensorFlow >= 2.1 code:

history = model.fit(training_set.repeat(),
                    steps_per_epoch=int(8000/batch_size),
                    epochs=25,
                    validation_data=test_set.repeat(),
                    validation_steps=int(2000/batch_size))

Notice that int(8000/batch_size) is equivalent to 8000 // batch_size (integer division)

============================================================================

也就是steps_per_epoch=int(8000/batch_size)，这里的8000是训练样本数

4、实例

训练样本为1000张，

train_datagen = ImageDataGenerator(     
    rescale=1./255,     
    rotation_range=40,     
    width_shift_range=0.2,     
    height_shift_range=0.2, 
    shear_range=0.2,     
    zoom_range=0.2,     
    horizontal_flip=True,) 
# 注意，不能增强验证数据
test_datagen = ImageDataGenerator(rescale=1./255) 


# 这里batch_size不能是32，不然就报如下错误
'''
WARNING:tensorflow:Your input ran out of data; 
interrupting training. Make sure that your dataset or generator can generate at least 
steps_per_epoch * epochs batches (in this case, 5000 batches). 
You may need to use the repeat() function when building your dataset.

'''
# 可能是整除关系吧



train_generator = train_datagen.flow_from_directory(         
    train_dir, # 目标目录         
    target_size=(150, 150), # 将所有图像的大小调整为 150×150
    batch_size=20, # 因为使用了 binary_crossentropy 损失，所以需要用二进制标签
    class_mode='binary') 

validation_generator = test_datagen.flow_from_directory(         
    validation_dir,         
    target_size=(150, 150),         
    batch_size=20,         
    class_mode='binary')

history = model.fit(       
    train_generator,
    steps_per_epoch=100,
    epochs=150,
    validation_data=validation_generator,
    validation_steps=50)

如果上面的batch_size=32，那么这里如果steps_per_epoch=100会报错

steps_per_epoch 参数的作用：从生成器中抽取 steps_per_epoch 个批量后（即运行了 steps_per_epoch 次梯度下降），拟合过程将进入下一个轮次。

4.1、具体测试情况

# case 1
# 如果上面train_generator的batch_size是32，如果这里steps_per_epoch=100，那么会报错
"""
tensorflow:Your input ran out of data; interrupting training.
Make sure that your dataset or generator can generate at least `steps_per_epoch * epochs` batches (in this case, 50 batches).
You may need to use the repeat() function when building your dataset.
"""
# 因为train样本数是2000（猫1000，狗1000），小于100*32
# case 2
# 如果上面train_generator的batch_size是20，如果这里steps_per_epoch=100，那么不会报错
# 因为大小刚好
# case 3
# 如果上面train_generator的batch_size是32，如果这里steps_per_epoch=int(1000/32)，
# 那么不会报错，但是会有警告，因为也是不整除
# 不会报错因为int(1000/32)*32 < 2000
# case 4
# 如果上面train_generator的batch_size是40，如果这里steps_per_epoch=100，照样报错
# 因为40*100>2000