如何从字典值返回 object？

Question

I have several classes, each holding its own Neural Network architecture.

Based on a user-entered flag, rx_flag , I am trying to retrieve a specific architecture in the driver file.

I have two problems:

1. I am unable to create a dictionary in the usual way. The following format is not working:

    def build_model(rx_flag): switcher = { 'xss': XSS().get_model(), 'rss': RSS().get_model() } return switcher.get(rx_flag)

2. After some research, I am finally able to at least construct a dictionary where the models in those classes are stored as values, but when I return them to the main() method, I get the NoneType .

Here is my class. The other classes have a similar template. I have commented out the implementation of __hash__() and __eq__() because storing in the dictionary seems to be working fine without it too.

from model import Model
from keras.layers import Dense
from keras.models import Sequential

class XSS(Model):

    def __init__(self):
        self.num_layers = 2
        self.input_dim = 3
        self.output_dim = 1
        self.architecture = [64, 32]
        self.model = Sequential()

    def get_model( self , arch=[64, 32]):
        # add input layer
        self.model.add(Dense(arch[0], activation='relu', input_shape=(self.input_dim, )))

        # add intermediate layers
        for i in range(1, self.num_layers):
            self.model.add(Dense(arch[i], activation='relu'))

        # add output layer
        self.model.add(Dense(self.output_dim, activation='linear'))
        return self.model

    def get_name( self ):
        return 'xss'

    def get_value( self ):
        return self.__value()

    def __value( self ):
        return (self.model, self.num_layers, self.input_dim, self.output_dim, self.architecture)

    # def __hash__(self):
    #     return (self.hash(self.__value()))
    #
    # def __eq__(self, other):
    #     if isinstance(other, XSS):
    #         return self.__value() == other.__value()
    #     return NotImplemented

This is the driver code:

import sys

from model import Model
from xss import XSS

def build_model(rx_flag):
    switcher = {}
    obX = XSS()
    switcher[obX.get_name()] = obX.get_model()
    obR = RSS()
    switcher[obR.get_name()] = obR.get_model()
    print(switcher)
    return switcher.get(rx_flag)

if __name__ == '__main__':
    rx_flag = sys.argv[0]
    # create a model instance based on flag
    model = build_model(rx_flag)
    model.summary()

This is the error that I get on attempting model.summary() .

Traceback (most recent call last):
File "C:/Users/path/driver.py", line 19, in <module>
    model.summary()
AttributeError: 'NoneType' object has no attribute 'summary'

How can I build the dictionary in a more Pythonic way, and have it return the actual model?

Answer 1

@juanpa.arrivillaga suggested that rx_flag is not what I think it is. They are right.

The code works fine even on PyCharm when I initialize rx_flag as below:

rx_flag = sys.argv[1]

I was under the impression that it is the first parameter when I enter it in the run configuration on PyCharm.