def __init__(self, inp_size, hidden_size_1, hidden_size_2, hidden_size_3, output_size): super(network, self).__init__() self.gru_1 = nn.GRU(inp_size, hidden_size_1) # for the conditioning weights weights self.cond_gauss = gv.conditional_gaussians self.cond_linear = nn.Linear(hidden_size_1, self.cond_gauss * 3) self.conditioner = cm.conditioner() self.condition_len = len(self.conditioner.string) self.gru_2 = nn.GRU(hidden_size_1 + self.condition_len, hidden_size_2) self.linear_1 = nn.Linear(hidden_size_2 + self.condition_len, hidden_size_3) self.linear_2 = nn.Linear( hidden_size_3 + hidden_size_2 + hidden_size_1 + self.condition_len, output_size) self.inp_size = inp_size self.hidden_size_1 = hidden_size_1 self.hidden_size_2 = hidden_size_2 self.hidden_size_3 = hidden_size_3 self.output_size = output_size # for training self.drop = nn.Dropout(0.4) ## not being used as its not overfitting to train self.relu = nn.ReLU() self.go = graves_output.network()
def __init__(self, inp_size, hidden_size_1, output_size): super(network, self).__init__() self.gru_1 = nn.GRU(inp_size, hidden_size_1) self.linear = nn.Linear(hidden_size_1, output_size) self.hidden_size_1 = hidden_size_1 self.inp_size = inp_size # for training self.drop = nn.Dropout(0.4) self.relu = nn.ReLU() self.go = graves_output.network()
def __init__(self, inp_size, hidden_size_1,hidden_size_2,hidden_size_3,output_size): super(network, self).__init__() self.gru_1 = nn.GRU(inp_size, hidden_size_1) self.gru_2 = nn.GRU(hidden_size_1+inp_size, hidden_size_2) self.linear_1 = nn.Linear(hidden_size_2+inp_size,hidden_size_3) self.linear_2 = nn.Linear(hidden_size_3+hidden_size_2+hidden_size_1,output_size) self.inp_size = inp_size self.hidden_size_1 = hidden_size_1 self.hidden_size_2 = hidden_size_2 self.hidden_size_3 = hidden_size_3 self.output_size = output_size # for training self.drop = nn.Dropout(0.4) ## not being used as its not overfitting to train self.relu = nn.ReLU() self.go = graves_output.network()