local variable 'opt_Learning_Rate2' referenced before assignment

[ShawnBio]

Hi, my codes encounter a bug:
local variable 'opt_Learning_Rate2' referenced before assignment
My codes are below.

from GENetLib.sim_data import sim_data_scalar
from GENetLib.grid_scalar_ge import grid_scalar_ge
ytype = 'Survival'
num_hidden_layers = 2
nodes_hidden_layer = [500, 1000]
learning_rate2 = [0.035, 0.045]
Lambda = [0.1]
learning_rate1 = [0.04, 0.05]
lambda2 = [0.07, 0.09]
num_epochs = 100
scalar_survival_linear = sim_data_scalar(rho_G = 0.25, rho_E = 0.3, dim_G = 500, dim_E = 5, n = 1500, dim_E_Sparse = 2, ytype = 'Survival', n_inter = 30)
y = scalar_survival_linear['y']
G = scalar_survival_linear['G']
E = scalar_survival_linear['E']
grid_scalar_ge_res = grid_scalar_ge(y, G, E, ytype, num_hidden_layers, nodes_hidden_layer, num_epochs,
              learning_rate1, learning_rate2, lambda1 = None, lambda2 = lambda2, Lambda = Lambda, threshold = 0.05)

Where is the problem actually?

[Barry57]

Here, nodes_hidden_layer = [500, 1000] is incorrect. The deeper layer must contain more nodes. Try to use [1000,100].

[ShawnBio]

Thanks for your reply! I have fixed my bug. By the way , I wonder why bug reports local variable 'opt_Learning_Rate2' referenced before assignment?

[Barry57]

As our neural network training has two steps: first training without MCP to optimize the l2 parameters, then training the MCP parameters based on the optimal l2 parameters. This error shows there are no optimal l2 parameters, that is, the first step of training reported an error and there's no proper model.

[ShawnBio]

I see. Thanks again!