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