TypeError: Fetch argument has invalid type class ‘numpy.float32‘, must be a string or Tensor

项目github地址：bitcarmanlee easy-algorithm-interview-and-practice
欢迎大家star，留言，一起学习进步

1.问题

tensorflow代码在运行的时候，出现如下错误

Epoch: 0010 cost= 0.080483146 W= 0.23614137 b= 0.97142047
...
TypeError: Fetch argument 0.080483146 has invalid type <class 'numpy.float32'>, must be a string or Tensor. (Can not convert a float32 into a Tensor or Operation.)

对应部分的代码如下

 ...loss = tf.reduce_sum(tf.pow(pred-Y, 2) / (2*n_samples))# optmizeoptimizer = tf.train.GradientDescentOptimizer(learning_rate).minimize(loss)init = tf.global_variables_initializer()with tf.Session() as sess:sess.run(init)for epoch in range(training_epochs):for (x, y) in zip(train_X, train_Y):sess.run(optimizer, feed_dict={X: x, Y: y})if (epoch+1) % display_step == 0:loss = sess.run(loss, feed_dict={X: train_X, Y:train_Y})print("Epoch:", '%04d' % (epoch + 1), "cost=", "{:.9f}".format(loss), "W=", sess.run(W), "b=", sess.run(b))...

2.原因分析

很明显是loss函数那边出了问题。第一次loss有正常输出，而后面loss则有异常，原因是Fetch需要的参数为一个Tensor或者String，但是现在传过来的确实个float32。

再查看一下

    loss = tf.reduce_sum(tf.pow(pred-Y, 2) / (2*n_samples))

第一轮迭代的时候，这个loss是个Tensor。但是后面 run方法返回的，确实个float，所以会出现异常！

3.解决方法

因为是loss的异常引起的，所以我们将run方法返回的loss重新命令，给个新变量即可。

    loss = tf.reduce_sum(tf.pow(pred-Y, 2) / (2*n_samples))# optmizeoptimizer = tf.train.GradientDescentOptimizer(learning_rate).minimize(loss)init = tf.global_variables_initializer()with tf.Session() as sess:sess.run(init)for epoch in range(training_epochs):for (x, y) in zip(train_X, train_Y):sess.run(optimizer, feed_dict={X: x, Y: y})if (epoch+1) % display_step == 0:newloss = sess.run(loss, feed_dict={X: train_X, Y:train_Y})print("Epoch:", '%04d' % (epoch + 1), "cost=", "{:.9f}".format(newloss), "W=", sess.run(W), "b=", sess.run(b))

像上面这样，将run方法返回的值重新命名为newloss即可解决上述问题！

TypeError: Fetch argument has invalid type class ‘numpy.float32‘, must be a string or Tensor相关推荐

has invalid type class 'numpy.ndarray', must be a string or Tensor
has invalid type <class 'numpy.ndarray'>, must be a string or Tensor. (Can not convert a ndarr ...
gee错误：Image.multiply, argument ‘image2‘: Invalid type. Expected type: Image＜unknown bands＞.
我们在计算reduceRegion进行相应的线性统计,比如说 Image (Error) Image.multiply, argument 'image2': Invalid type. Expect ...
成功解决TypeError: ‘encoding’ is an invalid keyword argument for this function
成功解决TypeError: 'encoding' is an invalid keyword argument for this function 目录解决问题解决思路解决方法解决问题 Ty ...
Numpy出现TypeError: integer argument expected, got float的分析与解决
问题今天在使用Numpy的时候,打印array的时候突然出现了TypeError: integer argument expected, got float 这个报错,部分代码如下: import ...
TypeError: first argument must be an iterable of pandas objects, you passed an object of type “DataF
使用concat()函数拼接两个表格,出现以下错误: TypeError: first argument must be an iterable of pandas objects, you pass ...
讲解为什么会提示“invalid type argument of ‘-＞‘ (have ‘Data_Ao‘)”错误信息
今天分享一个之前遇到的错误提示,其实就是语法的错误.可能如果对"->"或者"."的用法不了解,就会容易造成这样的错误,加上如果你的编译器不会提示一下基础的 ...
TypeError: mean() received an invalid combination of arguments - got (list, dim=int), but expected o
TypeError: mean() received an invalid combination of arguments - got (list, dim=int), but expected o ...
TypeError: Cannot handle this data type: (1, 1, 28), |u1
PIL和numpy相互转换方式如下: from PIL import Image#PIL image转成numpy np_img = np.asarray(PIL_img) #参数时PIL类型的图片 ...
{TypeError}clamp(): argument 'min' must be Number, not Tensor
{TypeError}clamp(): argument 'min' must be Number, not Tensor x1是二维数组, xx1 = x1[order[1:]].clamp(min ...
报错解决——ctypes.ArgumentError: argument 1:……….. : wrong type
运行 python darknet.py 结果报错如下: Traceback (most recent call last): File "darknet.py", line 13 ...

TypeError: Fetch argument has invalid type class ‘numpy.float32‘, must be a string or Tensor

1.问题

2.原因分析

3.解决方法

TypeError: Fetch argument has invalid type class ‘numpy.float32‘, must be a string or Tensor相关推荐

最新文章

热门文章