number_of_features=number_of_features_I, hidden_size=hidden_size, hidden_layer_depth=hidden_layer_depth, latent_length=latent_length, batch_size=batch_size, learning_rate=learning_rate, n_epochs=n_epochs, dropout_rate=dropout_rate, cuda=cuda, model_name=model_name_I, header=header_I, device=device) model_I.to(device) # Initialize training settings optimB = optim.Adam(model_B.parameters(), lr=learning_rate) optimI = optim.Adam(model_I.parameters(), lr=learning_rate) cl_loss_fn = nn.NLLLoss() recon_loss_fn = nn.MSELoss() # one stage training: with recon_loss and mse_loss training_start = datetime.now() # create empty lists to fill stats later epoch_train_loss_B = [] epoch_train_acc_B = [] epoch_val_loss_B = [] epoch_val_acc_B = [] max_val_acc_B = 0 epoch_train_loss_I = []
print_every=print_every, clip=clip, max_grad_norm=max_grad_norm, dload = logDir, model_name=model_name, header=header, device = device) model.to(device) # In[7]: criterion = nn.CrossEntropyLoss() optimizer = optim.Adam(model.parameters(), lr=learning_rate) cl_loss_fn = nn.NLLLoss() recon_loss_fn = nn.MSELoss() # In[8]: # 1st stage training: with recon_loss training_start=datetime.now() #split fit epoch_train_loss = [] epoch_train_acc = [] epoch_val_loss = [] epoch_val_acc = [] max_val_acc = 0