类型错误：__init__() 为参数“strides”获得了多个值

Question

The Nvidia model is showing error for strides, even if I initialize them to the default value of (1,1) I am using 'strides' as a replacement for the 'subsample' argument in previous versions of keras can someone explain the new syntax of using them.

def nvidia_model():
  model = Sequential()
  model.add(Conv2D(24,5,5, strides = (2,2), input_shape= (66,200,3), activation='relu'))
  model.add(Conv2D(36,5,5, strides = (2,2), activation = 'relu'))
  model.add(Conv2D(48,5,5, strides = (2,2), activation = 'relu'))
  model.add(Conv2D(64,3,3, activation = 'relu'))
  model.add(Conv2D(64,3,3, activation = 'relu'))
  model.add(Dropout(0.5))                                               #50% nodes turned to zero

  model.add( Flatten())
  model.add(Dense(100, activation ='relu'))
  model.add(Dropout(0.5))

  model.add(Dense(50, activation ='relu'))
  model.add(Dense(10, activation ='relu'))
  model.add(Dense(1))
  
  optimizer = Adam(lr = 1e-3)
  model.compile(loss = 'mse' , optimizer = optimizer)
  return model

model = nvidia_model()
print(model.summary)

Error:
TypeError                                 Traceback (most recent call last)
<ipython-input-83-aff2a2709b79> in <module>()
----> 1 model = nvidia_model()
      2 print(model.summary)

<ipython-input-82-7942ade664af> in nvidia_model()
      1 def nvidia_model():
      2   model = Sequential()
----> 3   model.add(Conv2D(24,5,5, strides = (2,2), input_shape= (66,200,3), activation='relu'))
      4   model.add(Conv2D(36,5,5, strides = (2,2), activation = 'relu'))
      5   model.add(Conv2D(48,5,5, strides = (2,2), activation = 'relu'))

TypeError: __init__() got multiple values for argument 'strides'

Answer 1

I never worked with tensorflow , but according to the documentation of Conv2D it's defined as

tf.keras.layers.Conv2D(
    filters, kernel_size, strides=(1, 1), padding='valid', data_format=None,
    dilation_rate=(1, 1), groups=1, activation=None, use_bias=True,
    kernel_initializer='glorot_uniform', bias_initializer='zeros',
    kernel_regularizer=None, bias_regularizer=None, activity_regularizer=None,
    kernel_constraint=None, bias_constraint=None, **kwargs
)

As you can see strides is the third parameter.

Now you use Conv2D(24,5,5, strides = (2,2), input_shape= (66,200,3), activation='relu') where the third parameter is 5 and then you try to set strides again with the keyword parameter. It seems there is one parameter too much in your call.

Imagine the little Python gnome handling your code: "OK, the boss wants an instance of Conv2D . He sets the first argument filters to 4 , the second argument kernel_size to 2 and the third argument strides to 2 . Done with the positional arguments. Now let's continue with the keyword arguments. Here we have strides and ... oh, I already have strides , so I don't know what to do. The boss might be angry so I'll tell him exactly what happenend: TypeError: __init__() got multiple values for argument 'strides' "

Answer 2

model.add(Conv2D(24,5,5, strides = (2,2), input_shape= (66,200,3), activation='relu'))
model.add(Conv2D(36,5,5, strides = (2,2), activation = 'relu'))
model.add(Conv2D(48,5,5, strides = (2,2), activation = 'relu'))

For these lines use parentheses like the following example

model.add(Conv2D(24,(5,5), strides = (2,2), input_shape= (66,200,3), activation='relu'))
model.add(Conv2D(36,(5,5), strides = (2,2), activation = 'relu'))
model.add(Conv2D(48,(5,5), strides = (2,2), activation = 'relu'))