无监督学习中的 NaN 损失函数

Question

我正在使用 conv1d 模型处理无监督数据集，如下所示

X = np.genfromtxt("C:/Users/hp/Desktop/test.txt")
X = np.expand_dims(X, axis=2)
model = Sequential()
model.add(Conv1D(12, 3, input_shape=(6621,1),
                 padding='same', strides=1, activation='relu',
                 kernel_regularizer=l2(0.01)))
model.add(Dropout(0.1))
model.add(Conv1D(15, 3, padding='same', strides=1, activation='relu',
                 kernel_regularizer=l2(0.01)))
model.add(Dropout(0.2))
model.add(Conv1D(118, 3, padding='same', strides=1, activation='relu',
                 kernel_regularizer=l2(0.01)))
model.add(Dropout(0.3))
##model.add(Flatten())
model.add(Dense(128, activation='relu', kernel_regularizer=l2(0.01)))
model.add(Dropout(0.3))
model.add(Dense(64, activation='relu', kernel_regularizer=l2(0.01)))
model.add(Dropout(0.3))
model.add(Dense(32, activation='relu', kernel_regularizer=l2(0.01)))
model.add(Dropout(0.3))
model.add(Dense(1, activation='sigmoid', kernel_regularizer=l2(0.01)))
print(model.summary())
model.compile(loss='kld',
              metrics=['accuracy'], optimizer='sgd')

model.fit(X, X, batch_size=128, verbose = 2, epochs=5, validation_split=0.2)

我尝试了不同的损失和优化器，但输出仍然让我损失了 NAN 并且准确性也很奇怪

<块引用>

122s - 损失：nan - 准确度：0.0896 - val_loss：nan - val_accuracy： 0.0000e+00

模型摘要是

    _________________________________________________________________
Layer (type)                 Output Shape              Param #   
=================================================================
conv1d_1 (Conv1D)            (None, 6621, 12)          48        
_________________________________________________________________
dropout_1 (Dropout)          (None, 6621, 12)          0         
_________________________________________________________________
conv1d_2 (Conv1D)            (None, 6621, 15)          555       
_________________________________________________________________
dropout_2 (Dropout)          (None, 6621, 15)          0         
_________________________________________________________________
conv1d_3 (Conv1D)            (None, 6621, 118)         5428      
_________________________________________________________________
dropout_3 (Dropout)          (None, 6621, 118)         0         
_________________________________________________________________
dense_1 (Dense)              (None, 6621, 128)         15232     
_________________________________________________________________
dropout_4 (Dropout)          (None, 6621, 128)         0         
_________________________________________________________________
dense_2 (Dense)              (None, 6621, 64)          8256      
_________________________________________________________________
dropout_5 (Dropout)          (None, 6621, 64)          0         
_________________________________________________________________
dense_3 (Dense)              (None, 6621, 32)          2080      
_________________________________________________________________
dropout_6 (Dropout)          (None, 6621, 32)          0         
_________________________________________________________________
dense_4 (Dense)              (None, 6621, 1)           33        
=================================================================
Total params: 31,632
Trainable params: 31,632
Non-trainable params: 0
_________________________________________________________________
None