• InvalidArgumentError：无法将 dtype 资源的张量转换为 NumPy 数组（无法将位于的资源添加到web应用程序的缓存中）
• cut 函数：无法根据规则“安全”将数组数据从 dtype('float64') 转换为 dtype('
• InvalidArgumentError: 预期 'tf.Tensor(False, shape=(), dtype=bool)' 为真
• InvalidArgumentError: 预期 'tf.Tensor(False, shape=(), dtype=bool)' 为真汇总数据：b'没有文件匹配模式：
• InvalidArgumentError：您必须使用 dtype float 和 shape [?,1,19,1] Keras 为占位符张量“conv2d_17_input”提供一个值

InvalidArgumentError：无法将 dtype 资源的张量转换为 NumPy 数组（无法将位于的资源添加到web应用程序的缓存中）

如何解决InvalidArgumentError：无法将 dtype 资源的张量转换为 NumPy 数组

我想使用 tf-hub 构建文本分类模型并导出为 tflite 模型但是
在转换包括 tf hub 的 tensorflow 模型时，我遇到了错误。请帮我解决。

import tensorflow as tf
import tensorflow_hub as hub 
model = tf.keras.Sequential()
model.add(tf.keras.layers.InputLayer(dtype=tf.string,input_shape=()))
model.add(hub.KerasLayer("https://tfhub.dev/google/tf2-preview/nnlm-en-dim50/1"))
converter=tf.lite.TFLiteConverter.from_keras_model(model)
tflite_model = converter.convert()

我尝试了 tf-lite python 和命令行 api。但是我遇到了 InvalidArgumentError。

InvalidArgumentError                      Traceback (most recent call last)
<ipython-input-15-5a8dbd778645> in <module>()
      5 model.add(hub.KerasLayer("https://tfhub.dev/google/tf2-preview/nnlm-en-dim50/1"))
      6 converter = tf.lite.TFLiteConverter.from_keras_model(model)
----> 7 tflite_model = converter.convert()
6 frames
/usr/local/lib/python3.6/dist-packages/tensorflow/lite/python/lite.py in convert(self)
    850     frozen_func,graph_def = (
    851         _convert_to_constants.convert_variables_to_constants_v2_as_graph(
--> 852             self._funcs[0],lower_control_flow=False))
    853 
    854     input_tensors = [
/usr/local/lib/python3.6/dist-packages/tensorflow/python/framework/convert_to_constants.py in convert_variables_to_constants_v2_as_graph(func,lower_control_flow,aggressive_inlining)
   1103       func=func,1104       lower_control_flow=lower_control_flow,-> 1105       aggressive_inlining=aggressive_inlining)
   1106 
   1107   output_graph_def,converted_input_indices = _replace_variables_by_constants(
/usr/local/lib/python3.6/dist-packages/tensorflow/python/framework/convert_to_constants.py in __init__(self,func,aggressive_inlining,variable_names_allowlist,variable_names_denylist)
    804         variable_names_allowlist=variable_names_allowlist,805         variable_names_denylist=variable_names_denylist)
--> 806     self._build_tensor_data()
    807 
    808   def _build_tensor_data(self):
/usr/local/lib/python3.6/dist-packages/tensorflow/python/framework/convert_to_constants.py in _build_tensor_data(self)
    823         data = map_index_to_variable[idx].numpy()
    824       else:
--> 825         data = val_tensor.numpy()
    826       self._tensor_data[tensor_name] = _TensorData(
    827           numpy=data,/usr/local/lib/python3.6/dist-packages/tensorflow/python/framework/ops.py in numpy(self)
   1069     """
   1070     # Todo(slebedev): Consider avoiding a copy for non-cpu or remote tensors.
-> 1071     maybe_arr = self._numpy()  # pylint: disable=protected-access
   1072     return maybe_arr.copy() if isinstance(maybe_arr,np.ndarray) else maybe_arr
   1073 
/usr/local/lib/python3.6/dist-packages/tensorflow/python/framework/ops.py in _numpy(self)
   1037       return self._numpy_internal()
   1038     except core._NotOkStatusException as e:  # pylint: disable=protected-access
-> 1039       six.raise_from(core._status_to_exception(e.code,e.message),None)  # pylint: disable=protected-access
   1040 
   1041   @property
/usr/local/lib/python3.6/dist-packages/six.py in raise_from(value,from_value)
InvalidArgumentError: Cannot convert a Tensor of dtype resource to a NumPy array.

解决方法

cut 函数：无法根据规则“安全”将数组数据从 dtype('float64') 转换为 dtype('

如何解决cut 函数：无法根据规则“安全”将数组数据从 dtype(''float64'') 转换为 dtype(''<U32'')

我想将 Dataframe 中某一列的内容更改为“好”或“坏”。 该列填充了从 1 到 10 的数字。 1-5 是坏的，6-10 是好的。 为此，我想使用 cut 方法。

bins = (1,5.5,10)
rating = [''bad'',''good'']
game[''useropinion''] = pd.cut(rating,bins)

运行后的结果：

Cannot cast array data from dtype(''float64'') to dtype(''<U32'') according to the rule ''safe''

怎么了？我该如何解决？

解决方法

你可以这样做：

game[''useropinion''] = pd.cut(game[''useropinion''],bins,labels=rating)

编辑： 为了回答问题，您试图削减评级而不是用户意见数据，因此很自然地会得到 TypeError，因为评级是一个字符串数组，而您的 bin 是数字。 ''

InvalidArgumentError: 预期 'tf.Tensor(False, shape=(), dtype=bool)' 为真

如何解决InvalidArgumentError: 预期 ''tf.Tensor(False, shape=(), dtype=bool)'' 为真

在使用结构相似性指数比较图像之前，我使用 PCA 来减少图像的尺寸。使用PCA后，tf.image.ssim报错。

我在这里比较图像而不使用 PCA。这完美地工作 -

import numpy as np
import tensorflow as tf
import time
(x_train,y_train),(x_test,y_test) = tf.keras.datasets.mnist.load_data(
    path=''mnist.npz''
)
start = time.time()
for i in range(1,6000):
    x_train_zero = np.expand_dims(x_train[0],axis=2)
    x_train_expanded = np.expand_dims(x_train[i],axis=2)
    print(tf.image.ssim(x_train_zero,x_train_expanded,255))
print(time.time()-start)

我在这里应用了 PCA 来减少图像的尺寸，以便 SSIM 花费更少的时间来比较图像 -

from sklearn.preprocessing import StandardScaler
from sklearn.decomposition import PCA
x_train = x_train.reshape(60000,-1)
scaler = StandardScaler()
X_scaled = scaler.fit_transform(x_train)
pca = PCA()
pca = PCA(n_components = 11)
X_pca = pca.fit_transform(X_scaled).reshape(60000,11,1)
start = time.time()
for i in range(1,6000):
    X_pca_zero = np.expand_dims(X_pca[0],axis=2)
    X_pca_expanded = np.expand_dims(X_pca[i],axis=2)
    print(tf.image.ssim(X_pca_zero,X_pca_expanded,255))
print(time.time()-start)

这段代码抛出错误 - InvalidArgumentError: Expected ''tf.Tensor(False,shape=(),dtype=bool)'' to be true。汇总数据：11,1,1 11

解决方法

所以，简而言之，发生该错误是因为在 cancel 中，输入 tf.image.ssim 和 X_pca_zero 的大小与 X_pca_expanded 不匹配，如果您有 {{1} } 那么 filter_size 和 filter_size=11 必须至少为 11x11，您可以如何更改代码的示例：

X_pca_zero

InvalidArgumentError: 预期 'tf.Tensor(False, shape=(), dtype=bool)' 为真汇总数据：b'没有文件匹配模式：

如何解决InvalidArgumentError: 预期 ''tf.Tensor(False, shape=(), dtype=bool)'' 为真汇总数据：b''没有文件匹配模式：

当我在 Google colab 上运行以下代码时，

table(selection)

我收到以下错误：

tf.data.Dataset.list_files(''/content/gdrive/MyDrive/Experiment/train/*.jpg'')

过去两周我一直被这个问题困扰，请帮帮我。另外，在运行上述代码行之前，我已经成功安装了 Google 驱动器。

解决方法

这是我正在使用的示例。

from google.colab import drive
drive.mount(''/content/drive'')
def load_image(filepath):
    raw_img = tf.io.read_file(filepath) 
    img_tensor_int = tf.image.decode_jpeg(raw_img,channels=3)
    img_tensor_flt = tf.image.convert_image_dtype(img_tensor_int,tf.float32)
    return img_tensor_flt,img_tensor_flt
def load_dataset(split):
    print(''/content/drive/MyDrive/CelebAsubset/''+split+''/*.jpg'')
    train_list_ds = tf.data.Dataset.list_files(''/content/drive/MyDrive/CelebAsubset/''+split+''/*.jpg'',shuffle=False)
    train_ds = train_list_ds.map(load_image)
    return train_ds
train_ds = load_dataset(''train'')
val_ds = load_dataset(''val'')
test_ds = load_dataset(''test'')

InvalidArgumentError：您必须使用 dtype float 和 shape [?,1,19,1] Keras 为占位符张量“conv2d_17_input”提供一个值

如何解决InvalidArgumentError：您必须使用 dtype float 和 shape [?,1,19,1] Keras 为占位符张量“conv2d_17_input”提供一个值

我正在尝试使用 Keras 构建 GAN 模型，但遇到了一些错误消息的问题，InvalidArgumentError: You must Feed a value for placeholder tensor ''conv2d_17_input'' with dtype float and shape [?,1,19,1 ].

数据来自真实数据，并试图用它解决不平衡问题。这就是数据的形状与 32x32 等普通图像不同的原因。

如果有人知道这里发生了什么，请帮助我。真的很感激。

代码如下：

class Data:
    """
    Define dataset for training GAN
    """
    def __init__(self,data,batch_size,z_input_dim):
        X_train,y_train,X_test,y_test = train_test_split(data.iloc[:,:-1].values,data.result.values,test_size=0.3)
        self.x_data = X_train
        self.x_data = self.x_data.reshape((self.x_data.shape[0],1) + (self.x_data.shape[1],1))
        self.batch_size = batch_size
        self.z_input_dim = z_input_dim
 
    def get_real_sample(self):
        """
        get real sample mnist images
        :return: batch_size number of mnist image data
        """
        return self.x_data[np.random.randint(0,self.x_data.shape[0],size=self.batch_size)]
    
    def get_z_sample(self,sample_size):
        """
        get z sample data
        :return: random z data (batch_size,z_input_dim) size
        """
        return np.random.uniform(-1.0,1.0,(sample_size,self.z_input_dim))

和 GAN 模型：

class GAN:
    def __init__(self,learning_rate,z_input_dim):
        """
        init params
        :param learning_rate: learning rate of optimizer
        :param z_input_dim: input dim of z
        """
        self.learning_rate = learning_rate
        self.z_input_dim = z_input_dim
        self.D = self.discriminator()
        self.G = self.generator()
        self.GD = self.combined()
    def discriminator(self):
        """
        define discriminator
        """
        D = Sequential()
        D.add(Conv2D(128,(1,3),input_shape=(1,1),kernel_initializer=initializers.Randomnormal(stddev=0.02),data_format=''channels_last''))
        D.add(LeakyReLU(0.2))
        D.add(MaxPooling2D(pool_size=(1,2),strides=2,data_format=''channels_last''))
        D.add(Conv2D(256,padding=''same'',data_format=''channels_last''))
        D.add(Flatten())
        D.add(Dense(128))
        D.add(LeakyReLU(0.2))
        D.add(Dropout(0.3))
        D.add(Dense(1,activation=''sigmoid''))
        adam = Adam(lr=self.learning_rate,beta_1=0.5)
        D.compile(loss=''binary_crossentropy'',optimizer=adam,metrics=[''accuracy''])
        return D
    def generator(self):
        """
        define generator
        """
        G = Sequential()
        G.add(Dense(256,input_dim=self.z_input_dim))
        G.add(LeakyReLU(0.2))
        G.add(Dense(19))
        G.add(LeakyReLU(0.2))
        G.add(Batchnormalization())
        
        G.add(Reshape((1,input_shape = (19,)))
        G.add(Conv2D(1,activation=''tanh'',data_format=''channels_last''))
      
        adam = Adam(lr=self.learning_rate,beta_1=0.5)
        G.compile(loss=''binary_crossentropy'',metrics=[''accuracy''])
        return G
    def combined(self):
        """
        defien combined gan model
        """
        G,D = self.G,self.D
        D.trainable = False
        GD = Sequential()
        GD.add(G)
        GD.add(D)
        adam = Adam(lr=self.learning_rate,beta_1=0.5)
        GD.compile(loss=''binary_crossentropy'',metrics=[''accuracy''])
        D.trainable = True
        return GD

最后是操作代码：

class Model:
    def __init__(self,epochs,z_input_dim,n_iter_D,n_iter_G):
        self.epochs = epochs
        self.batch_size = batch_size
        self.learning_rate = learning_rate
        self.z_input_dim = z_input_dim
        self.data = Data(data,self.batch_size,self.z_input_dim)
        # the reason why D,G differ in iter : Generator needs more training than discriminator
        self.n_iter_D = n_iter_D
        self.n_iter_G = n_iter_G
        self.gan = GAN(self.learning_rate,self.z_input_dim)
        # print status
        batch_count = self.data.x_data.shape[0] / self.batch_size
        print(''Epochs:'',self.epochs)
        print(''Batch size:'',self.batch_size)
        print(''Batches per epoch:'',batch_count)
        print(''Learning rate:'',self.learning_rate)
        print(''Image data format:'',K.image_data_format())
    def fit(self):
        self.d_loss = []
        self.g_loss = []
        for epoch in range(self.epochs):
            # train discriminator by real data
            dloss = 0
            for iter in range(self.n_iter_D):
                dloss = self.train_D()
            # train GD by generated fake data
            gloss = 0
            for iter in range(self.n_iter_G):
                gloss = self.train_G()
            # save loss data
            self.d_loss.append(dloss)
            self.g_loss.append(gloss)
            # plot and save model each 20n epoch
            if epoch % 20 == 0:
                print("Epoch: {},discriminator loss: {},Generator loss: {}".format(str(epoch),str(dloss),str(gloss)))
        
        # show loss after train
        self.plot_loss_graph(self.g_loss,self.d_loss)
    def train_D(self):
        """
        train discriminator
        """
        # Real data
        real = self.data.get_real_sample()
        # Generated data
        z = self.data.get_z_sample(self.batch_size)
        generated_images = self.gan.G.predict(z)
        print(generated_images.shape)
        print(generated_images.dtype)
        
        # labeling and concat generated,real images
        x = np.concatenate((real,generated_images),axis=0)
        y = [0.9] * self.batch_size + [0] * self.batch_size
        # train discriminator
        self.gan.D.trainable = True
        loss = self.gan.D.train_on_batch(x,y)
        return loss
    def train_G(self):
        """
        train Generator
        """
        # Generated data
        z = self.data.get_z_sample(self.batch_size)
        # labeling
        y = [1] * self.batch_size
        # train generator
        self.gan.D.trainable = False
        loss = self.gan.GD.train_on_batch(z,y)
        return loss
    def plot_loss_graph(self,g_loss,d_loss):
        """
        Save training loss graph
        """
        # show loss graph
        plt.figure(figsize=(10,8))
        plt.plot(d_loss,label=''discriminator loss'')
        plt.plot(g_loss,label=''Generator loss'')
        plt.xlabel(''Epoch'')
        plt.ylabel(''Loss'')
        plt.legend()
        plt.show()

当我运行下面的代码时：

batch_size = 50
epochs = 100
learning_rate = 0.001
z_input_dim = 100
n_iter_D = 1
n_iter_G = 5
# run model
model = Model(data_dummise,n_iter_G)
model.fit()

我遇到了以下错误消息：

----------------------------------------------- ---------------------------- InvalidArgumentError 回溯（最近一次调用 最后）在 8#运行模型 9 模型 = 模型（data_dummise、batch_size、epochs、learning_rate、z_input_dim、n_iter_D、n_iter_G） ---> 10 model.fit()

合身（自我） 33 光泽 = 0 34 对于范围内的迭代（self.n_iter_G）： ---> 35 光泽 = self.train_G() 36 37 # 保存丢失数据

in train_G(self) 82#火车发电机 83 self.gan.D.trainable = False ---> 84 损失 = self.gan.GD.train_on_batch(z,y) 85回波损耗 86

~/.local/lib/python3.6/site-packages/tensorflow/python/keras/engine/training.py 在 train_on_batch(self,x,y,sample_weight,class_weight,第 1173 章 self._update_sample_weight_modes(sample_weights=sample_weights)
1174 self._make_train_function() -> 1175 输出 = self.train_function(ins) # pylint: disable=not-callable 1176 1177 if reset_metrics:

~/.local/lib/python3.6/site-packages/tensorflow/python/keras/backend.py 在调用（自我，输入）3290 3291 fetched = self._callable_fn(*array_vals,-> 3292 run_Metadata=self.run_Metadata) 3293
self._call_fetch_callbacks(fetched[-len(self._fetches):]) 3294
output_structure = nest.pack_sequence_as(

~/.local/lib/python3.6/site-packages/tensorflow/python/client/session.py 在 调用(self,*args,**kwargs) 1456 ret = tf_session.TF_Sessionruncallable(self._session._session,1457
self._handle,args,-> 1458 run_Metadata_ptr) 1459 如果 run_Metadata: 1460
proto_data = tf_session.TF_GetBuffer(run_Metadata_ptr)

InvalidArgumentError：您必须为占位符张量提供一个值 ''conv2d_17_input'' 带有 dtype 浮点数和形状 [?,1] [[{{node conv2d_17_input}}]]

关于InvalidArgumentError：无法将 dtype 资源的张量转换为 NumPy 数组和无法将位于的资源添加到web应用程序的缓存中的问题就给大家分享到这里，感谢你花时间阅读本站内容，更多关于cut 函数：无法根据规则“安全”将数组数据从 dtype('float64') 转换为 dtype('等相关知识的信息别忘了在本站进行查找喔。

本文标签：

• InvalidArgumentError
• Tensor_to_Numpy
• Web_Application_Caching
• Data_Type_Support