ValueError：检查模型目标时出错：传递给模型的Numpy数组列表不是模型期望的大小

Question

我有多路输出

out = [Dense(19, name='one', activation='softmax')(out),
           Dense(19, name='two', activation='softmax')(out),
           Dense(19, name='three', activation='softmax')(out),
           Dense(19, name='four', activation='softmax')(out)]


model.fit(reshape_train_X,  y_onehot, batch_size=400, epochs=100, verbose=2,
          validation_split=0.2, callbacks=callbacks_list)

这是我的y_onehot格式：

[array([[1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0],
       [0, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0],
       [0, 0, 0, 0, 0, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0],
       [0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 0, 0, 0, 0, 0, 0, 0]],
      dtype=uint8), array([[1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0],
       [0, 0, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0],
       [0, 0, 0, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0],
       [0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 0, 0, 0]],dtype=uint8),.....]

并且我收到此错误消息

ValueError: Error when checking model target: the list of Numpy arrays that you are passing to your model is not the size the model expected. Expected to see 4 array(s), but instead got the following list of 5000 arrays: [array([[1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0],
       [0, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0],
       [0, 0, 0, 0, 0, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0],
 ...

我不知道为什么y_onehot在数组中有四个列表时会发生此错误。

len（y_onehot）：5000

print（“ y_onehot”，y_onehot [0]）

[[1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0]
 [0 1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0]
 [0 0 0 0 0 1 0 0 0 0 0 0 0 0 0 0 0 0 0]
 [0 0 0 0 0 0 0 0 0 0 0 1 0 0 0 0 0 0 0]]

print（“ y_onehot”，len（y_onehot [0]））

y_onehot 4

我尝试this。但是仍然没有用。

感谢您的帮助。

Answer 1

这是一个虚拟的例子。注意你的y。您必须通过适合每个分离的输出

inp = Input((50))
x = Dense(32)(inp)
x1 = Dense(19, name='one', activation='softmax')(x)
x2 = Dense(19, name='two', activation='softmax')(x)
x3 = Dense(19, name='three', activation='softmax')(x)
x4 = Dense(19, name='four', activation='softmax')(x)

model = Model(inp, [x1,x2,x3,x4])
model.compile('adam', 'categorical_crossentropy')

X = np.random.uniform(0,1, (5000,50))
y1 = np.random.randint(0,2, (5000,19))
y2 = np.random.randint(0,2, (5000,19))
y3 = np.random.randint(0,2, (5000,19))
y4 = np.random.randint(0,2, (5000,19))

model.fit(X, [y1,y2,y3,y4], epochs=10)