ValueError：未为任何变量提供渐变-使用自定义损失函数和run_eagerly = True训练回归keras模型

如何解决ValueError：未为任何变量提供渐变-使用自定义损失函数和run_eagerly = True训练回归keras模型

我创建了一个虚拟回归keras模型，以在将其提供给实际模型之前检查我的自定义损失。 现在，我只希望运行此自定义损失函数。我想知道我在哪里以及为什么错了，以及如何解决自定义损失。
感谢您的帮助，并非常感谢。

Keras版本：2.2.4
Tensorflow版本：2.2
Python版本：3.7

[UPDATED_1]-我删除了损失函数中的所有numpy函数，但仍然遇到相同的错误。
这是我在[UPDATED_1]之后的代码：

import numpy as np
import keras
import tensorflow as tf
from keras.models import Sequential
from keras.layers import Dense
import keras.backend as K
import subprocess
#tf.config.experimental_run_functions_eagerly(True)
def my_loss(y_true,y_pred):
    y_true_shape = K.get_value(K.shape(y_true)[0])
    for i in range(y_true_shape):
        f = open('y_pred.txt','w')
        y_pred_temp = K.get_value(y_pred[i])
        f.write(str(y_pred_temp))
        f.close()
        f = open('y_true.txt','w')
        y_true_temp = K.get_value(y_true[i])
        f.write(str(y_true_temp))
        f.close()
        #temp = subprocess.call()# calls an outside program to read both y_true.txt and y_pred.txt 
        #and writes a loss value after being calculated into a txt file
        #bcs i'm testing this custom loss funtion so i just create a dummy txt file named "lossVals.txt" 
        #and its content is "1234"
        f = open('lossVals.txt','r')
        lossVal_data = f.read()
        f.close()
        lossVal_temp = K.variable((tf.strings.to_number(lossVal_data),))
        if i == 0:
            loss = lossVal_temp
        else:
            loss = K.concatenate((loss,lossVal_temp),axis=0)
    return loss
x_train = np.random.rand(1000,199)
y_train = np.random.rand(1000,199)
x_test = np.random.rand(200,199)
y_test = np.random.rand(200,199)
model = Sequential()
model.add(Dense(50,input_shape=(199,),activation='relu'))
model.add(Dense(20,activation='relu'))
model.add(Dense(10,activation='relu'))
model.add(Dense(199,activation='linear'))
model.compile(loss=my_loss,optimizer='adam',run_eagerly=True)
model.fit(x_train,y_train,batch_size=32,epochs=1)

这是错误日志

Traceback (most recent call last):

  File "<ipython-input-84-ebd97e65f38d>",line 1,in <module>
    model.fit(x_train,epochs=1)

  File "C:\Users\Deut\Anaconda3\envs\tensorflow\lib\site-packages\tensorflow\python\keras\engine\training.py",line 66,in _method_wrapper
    return method(self,*args,**kwargs)

  File "C:\Users\Deut\Anaconda3\envs\tensorflow\lib\site-packages\tensorflow\python\keras\engine\training.py",line 848,in fit
    tmp_logs = train_function(iterator)

  File "C:\Users\Deut\Anaconda3\envs\tensorflow\lib\site-packages\tensorflow\python\keras\engine\training.py",line 572,in train_function
    self.train_step,args=(data,))

  File "C:\Users\Deut\Anaconda3\envs\tensorflow\lib\site-packages\tensorflow\python\distribute\distribute_lib.py",line 951,in run
    return self._extended.call_for_each_replica(fn,args=args,kwargs=kwargs)

  File "C:\Users\Deut\Anaconda3\envs\tensorflow\lib\site-packages\tensorflow\python\distribute\distribute_lib.py",line 2290,in call_for_each_replica
    return self._call_for_each_replica(fn,args,kwargs)

  File "C:\Users\Deut\Anaconda3\envs\tensorflow\lib\site-packages\tensorflow\python\distribute\distribute_lib.py",line 2649,in _call_for_each_replica
    return fn(*args,**kwargs)

  File "C:\Users\Deut\Anaconda3\envs\tensorflow\lib\site-packages\tensorflow\python\autograph\impl\api.py",line 282,in wrapper
    return func(*args,line 541,in train_step
    self.trainable_variables)

  File "C:\Users\Deut\Anaconda3\envs\tensorflow\lib\site-packages\tensorflow\python\keras\engine\training.py",line 1804,in _minimize
    trainable_variables))

  File "C:\Users\Deut\Anaconda3\envs\tensorflow\lib\site-packages\tensorflow\python\keras\optimizer_v2\optimizer_v2.py",line 521,in _aggregate_gradients
    filtered_grads_and_vars = _filter_grads(grads_and_vars)

  File "C:\Users\Deut\Anaconda3\envs\tensorflow\lib\site-packages\tensorflow\python\keras\optimizer_v2\optimizer_v2.py",line 1219,in _filter_grads
    ([v.name for _,v in grads_and_vars],))

ValueError: No gradients provided for any variable: ['dense/kernel:0','dense/bias:0','dense_1/kernel:0','dense_1/bias:0','dense_2/kernel:0','dense_2/bias:0','dense_3/kernel:0','dense_3/bias:0'].