Regression accuracy for house price prediction is too high and not improving

Could somebody help me. I am new to Pytorch and am trying to predict house prices.
I have been able to batch my dataset and do a train-test-split. Even though I trained my model for many epochs my accuracy isn’t really improving or that good.

I used get an error of 1e10 (using MSE). Is this the best it can get or am I doing something wrong.

My training loop:

epochs=10
final_losses=[]
lrs=2
dif=3
for j in range(lrs):
  ###Backward Propogation-- Define the loss_function,define the optimizer
  loss_function=nn.MSELoss()
  optimizer=torch.optim.Adam(model.parameters(),lr=10**(-j-dif))
  print("\n\n","-"*12)
  print("Lr: ", str(10**(-j-dif)))
  print("-"*12)
  print("\n")
  for i in range(epochs):
    for step,(x, y) in enumerate(train):#tqdm
      y_pred=model(x)
      y_pred=y_pred
      loss=loss_function(y_pred,y)
      final_losses.append(loss)
      optimizer.zero_grad()
      loss.backward()
      optimizer.step()

    print("\nEpoch number: {} and the loss : {}".format(i+1,loss.item()))

Testing loop:

#### Prediction In X_test data
d=train
predictions=[]
for x,y in test:
  #print(x)!
  with torch.no_grad():
    for step,i in enumerate(range(len(x))):
      print("Count: {}".format(step+1))
      mse=abs(model(x[i]).tolist()[0]-y[i].tolist())**2
      print("Expected:",y[i].tolist()," Predicted: ",model(x[i]).tolist()[0]," Error: {:.2e}".format(mse))   
  break

My error plot:
image

My predictions:

I am confused why my expected vs predicted values differ so much.
My complete Notebook: https://colab.research.google.com/drive/1Vd9fJs0bfd7IZ8D5dHS859EThz1DEB58?usp=sharing
Any help would be highly appreciated.

I have a problem similar to your problem.
My code is working in keras but not in pytorch.
Please somebody help!