Question

使用youtube给出的原始模型，它在使用mnist时效果很好，但是我想转换为数据库。

我正在使用模型：

def base_model():
    model =Sequential()
    model.add(Dense(3, input_dim=3, kernel_initializer='normal',activation='relu'))
    model.add(Dense(second_layer, input_dim=3, kernel_initializer='normal',activation='relu'))
    model.add(Dense(third_layer, input_dim=second_layer, kernel_initializer='normal',activation='relu'))
    model.add(Dense(num_classes, kernel_initializer='normal',activation='softmax', name='preds'))

model = base_model()

model.summary()
model.fit(X_train,y_train, validation_data=(X_test,y_test), epochs=5,batch_size=100,verbose=2)

scores = model.evaluate(X_test,y_test,verbose=2)
print("Erro de : %.2f%%" % (100-scores[1]*100))
    model.compile(loss='categorical_crossentropy',optimizer='adam',metrics=['acc'])
    return model

我训练模型后，出现此错误：

ValueError：检查输入时出错：预期density_36_input具有 2维，但数组的形状为（374，1，3）

我想像下面这样转换一个数组：

array([[[0.3529358 , 0.32940674, 0.23529053]],

       [[0.32940674, 0.23529053, 0.11764526]],

       [[0.23529053, 0.11764526, 0.08235168]],

       ...,

       [[0.7529373 , 0.76470184, 0.7411728 ]],

       [[0.76470184, 0.7411728 , 0.7529373 ]],

       [[0.7411728 , 0.7529373 , 0.7411728 ]]], dtype=float32)


into this one:

array([[[0.3529358 , 0.32940674, 0.23529053],

       [0.32940674, 0.23529053, 0.11764526],

       [0.23529053, 0.11764526, 0.08235168]],

       ...,

       [[0.7529373 , 0.76470184, 0.7411728 ],

       [0.76470184, 0.7411728 , 0.7529373 ],

       [0.7411728 , 0.7529373 , 0.7411728 ]]], dtype=float32)

为什么我的阵列不能与训练模型一起使用？

0 个答案: