Python Tensorflow - InvalidArgumentError：无法计算 BatchMatMulV2 作为输入答案

【问题标题】：Python Tensorflow - InvalidArgumentError: cannot compute BatchMatMulV2 as inputPython Tensorflow - InvalidArgumentError：无法计算 BatchMatMulV2 作为输入
【发布时间】：2021-09-02 02:47:26
【问题描述】：

我是张量流的新手。我正在这样做-

def loss_function(y_pred, y_true):
    return -tf.reduce_mean(tf.matmul(y_true, tf.math.log(y_pred)) + tf.matmul(1-y_true, tf.math.log(1-y_pred)))

为此，我收到此错误-

---------------------------------------------------------------------------
InvalidArgumentError                      Traceback (most recent call last)
<ipython-input-28-84f43999b37a> in <module>()
     20   with tf.GradientTape() as tape:
     21     y_predicted = h(x_train,w)
---> 22     costF = loss_function(y_predicted, y)
     23 
     24   gradients = tape.gradient(costF, w)

5 frames
/usr/local/lib/python3.7/dist-packages/six.py in raise_from(value, from_value)

InvalidArgumentError: cannot compute BatchMatMulV2 as input #1(zero-based) was expected to be a int64 tensor but is a double tensor [Op:BatchMatMulV2]

有人可以帮忙吗？

【问题讨论】：

标签： python tensorflow tensor

【解决方案1】：

y_pred和y_true的dtype不同，只是转换之前的类型

def loss_function(y_pred, y_true):
    y_pred = tf.cast(y_pred, tf.float64)
    y_true = tf.cast(y_true, tf.float64)
    return -tf.reduce_mean(tf.matmul(y_true, tf.math.log(y_pred)) + tf.matmul(1-y_true, tf.math.log(1-y_pred)))

【讨论】：