cat_train = cats[:batch_size - test_size]
cat_test = cats[batch_size - test_size:batch_size]
con_train = conts[:batch_size - test_size]
con_test = conts[batch_size - test_size:batch_size]
y_train = y[:batch_size - test_size]
y_test = y[batch_size - test_size:batch_size]
# Neural Networky stuff
selfembeds = nn.ModuleList([nn.Embedding(ni, nf) for ni, nf in emb_szs])
torch.manual_seed(33)
model = TabularModel(emb_szs, conts.shape[1], 1, [200, 100], p=0.4)
criterion = nn.MSELoss() # we'll convert this to RMSE later
optimizer = torch.optim.Adam(model.parameters(), lr=0.001)
start_time = time.time()
epochs = 300
losses = []
for i in range(epochs):
i += 1
y_pred = model(cat_train, con_train)
loss = torch.sqrt(criterion(y_pred, y_train)) # RMSE
losses.append(loss)
# a neat trick to save screen space:
if i % 25 == 1:
print(f'epoch: {i} loss: {loss.item()}')
