Get parameters of modules in a dictionary

1

If a module contains a dictionary which has other two modules as following, can I get parameters of model_dict[‘model1’] and model_dict[‘model2’] with outer_network.state_dict()? Or is there any solutions to get parameters of modules from a dictionary?

class inner_network_1(nn.Module):
    def __init__(self):
        super(inner_network_1, self).__init__()
        self.inner_fc = nn.Linear(20, 30)

class inner_network_2(nn.Module):
    def __init__(self):
        super(inner_network_2, self).__init__()
        self.inner_fc = nn.Linear(20, 30)

class outer_network(nn.Module):
    def __init__(self):
        super(outer_network, self).__init__()
        self.fc = nn.Linear(10, 20)
        model_dict = {}
        model_dict['model1'] = inner_network_1()
        model_dict['model2'] = inner_network_2()
2

do you mean something like this,

class outer_network(nn.Module):
  def __init__(self):
    super().__init__()
    self.fc = nn.Linear(10, 20)
    self.inner_network_1 = nn.ModuleDict([['inner_fc', nn.Linear(20, 30)]])
    self.inner_network_2 = nn.ModuleDict([['inner_fc', nn.Linear(20, 30)]])

model = outer_network()
model.state_dict()

or something like this,

class inner_network_1(nn.Module):
    def __init__(self):
        super(inner_network_1, self).__init__()
        self.inner_fc = nn.Linear(20, 30)

class inner_network_2(nn.Module):
    def __init__(self):
        super(inner_network_2, self).__init__()
        self.inner_fc = nn.Linear(20, 30)

class outer_network(nn.ModuleDict):
    def __init__(self):
        super(outer_network, self).__init__()
        self.fc = nn.Linear(10, 20)
        self['model1'] = inner_network_1()
        self['model2'] = inner_network_2()

model = outer_network()
model.state_dict()
3

Thank you for these two examples! I didn’t know ModuleDict, which is exactly what I need!