TensorFlow——keras--报错+修改

1

mnist.load_data() 报错解决

#导入数据集
fashion_mnist = keras.datasets.fashion_mnist
(x_train_all, y_train_all),(x_test, y_test) = fashion_mnist.load_data()

2

Tensorflow2.0报错：ProfilerNotRunningError: Cannot stop profiling. No profiler is running.

#TensorBoard（文件夹）, EarlyStopping, ModelCheckpoint（文件名）
#logdir = './callbacks'  #定义文件夹
logdir = os.path.join("callbacks")
if not os.path.exists(logdir):
    os.mkdir(logdir)  #如果没有该文件，创建文件
output_model_file = os.path.join(logdir,
                                 "fashion_mnist_model.h5")  #输出的文件也放在该文件夹下
callbacks = [
    keras.callbacks.TensorBoard(logdir),
    keras.callbacks.ModelCheckpoint(output_model_file,
                                    save_best_only = True),  #保存最好的模型，否则最近的模型
    keras.callbacks.EarlyStopping(patience=5, min_delta=1e-3),
]

# EarlyStopping的参数： 
# monitor：关注的指标，一般是目标函数的值
# min_delta：预值，如果比预值低，就停
# patience：当前后两次目标函数的值比min_delta小多少次时，关掉

history = model.fit(x_train_scaled, y_train, epochs=10,
                    validation_data=(x_valid_scaled, y_valid),
                    callbacks = callbacks)
# epochs： 迭代10次 ;  validation：验证

修改

#logdir = './callbacks'  #定义文件夹
logdir = os.path.join("callbacks")

3

报错：AssertionError: Bad argument number for Name: 3, expecting 4
修改：

pip install gast==0.2.2  #0.3.2版本的问题，降级后可以

4

ValueError: Protocol message Features has no “features” field.

def serialize_example(x, y):
    """Converts x, y to tf.train.Example and serialize"""
    input_features = tf.train.FloatList(value = x)
    label = tf.train.FloatList(value = y)
    features = tf.train.Features(
        features = {
            "input_features": tf.train.Feature(
                float_list = input_features),
            "label": tf.train.Feature(float_list = label)
        }
    )
    example = tf.train.Example(features =features)
    return example.SerializeToString()

修改：取消features中的s

feature = {
            "input_features": tf.train.Feature(
                float_list = input_features),
            "label": tf.train.Feature(float_list = label)
        }

5

报错：inflate() failed with error -3: incorrect header check [Op:IteratorGetNextSync]

dataset = filename_dataset.interleave(
        lambda filename: tf.data.TFRecordDataset(
            filename, compression_type = "GZIP"),
        cycle_length = n_readers
    )

修改为：

dataset = dataset.interleave(
        lambda filename: tf.data.TFRecordDataset(
            filename, compression_type = "GZIP"),
        cycle_length = n_readers
    )